Swallowing simulation device and swallowing simulation method
The swallowing simulation device accurately models head and neck organ movements using muscle particles with specified fiber directions and contractile stress, improving the simulation of orally ingested products and diagnostic methods.
Patent Information
- Application Number
- JP2020189576
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2020-01-28
- Filing Date
- 2020-11-13
- Publication Date
- 2024-11-11
- Estimated Expiration
- 2040-11-13
AI Technical Summary
Existing swallowing simulation devices inadequately reproduce the movement of head and neck organs during swallowing, limiting the accuracy of developing products that prevent aspiration and effective diagnosis and treatment methods.
A swallowing simulation device and method that models head and neck organs using muscle particles with specified muscle fiber directions in a three-dimensional image, applying contractile stress based on muscle activity rates, and analyzes motion using a dynamic three-dimensional particle model to evaluate muscle activity rates accurately.
Enhances the accuracy of reproducing head and neck organ movements during swallowing, allowing for more precise simulation of orally ingested products and improved diagnostic and treatment methods.
Smart Images

Figure 0007580253000049 
Figure 0007580253000050 
Figure 0007580253000051
Abstract
Description
[Technical field]
[0001] The present invention relates to a swallowing simulation device and a swallowing simulation method. [Background technology]
[0002] The relationship between the food properties and the movements of the head and neck organs during swallowing is complex, and it is very difficult to accurately grasp the phenomenon itself. Swallowing is the movement of sending food (including beverages) taken into the oral cavity through the pharynx and esophagus to the stomach. During swallowing, the muscles of the oral cavity, pharynx, larynx, and esophagus act in a set sequence within a short period of time to perform complex movements.
[0003] Conventionally, a swallowing simulation device using a computer is known to simulate the behavior of a food bolus during swallowing (see, for example, Patent Document 1). In this swallowing simulation device, the movement of the oral organs and the physical properties of food and drink can be set, and the behavior of food and drink can be analyzed in a three-dimensional image using a particle method. Such a swallowing simulation device can approximately reproduce the actual phenomenon related to swallowing and visualize the swallowing phenomenon, and is thought to be useful, for example, in developing orally ingested foods, medicines, beverages, and other products that can suppress aspiration. [Prior art documents] [Patent documents]
[0004] [Patent Document 1] Patent No. 6022789 Summary of the Invention [Problem to be solved by the invention]
[0005] However, in order to develop oral intake products that can suppress aspiration or to establish effective diagnosis and treatment methods by elucidating the mechanism of aspiration, simple visualization of the swallowing phenomenon as shown in Patent Document 1 is insufficient; it is desirable to reproduce the movement of the head and neck organs when a simulated oral intake product is swallowed, as well as the behavior of the simulated oral intake product, more accurately than ever before.
[0006] The present invention has been made in consideration of the above points, and aims to provide a swallowing simulation device and a swallowing simulation method that can reproduce the movement of the head and neck organs during swallowing more accurately than conventional methods. [Means for solving the problem]
[0007] The swallowing simulation device according to the present invention includes a memory unit that stores a dynamic three-dimensional head and neck particle model in which a plurality of head and neck organs are modeled by a plurality of muscle particles having muscle fiber directions specified for each muscle type within a three-dimensional image, and in which a contraction stress based on the muscle fiber direction during swallowing is imparted to the muscle particles by a muscle activity rate; a muscle activity rate changing unit that changes the value of the muscle activity rate set for each muscle type; a motion analysis unit that uses the dynamic three-dimensional head and neck particle model in which the value of the muscle activity rate has been changed to analyze the movement of the head and neck organs during swallowing in the three-dimensional image based on a particle method; and a muscle activity rate analysis unit that calculates an evaluation function that evaluates the muscle activity rate based on the analysis results of the movement of the head and neck organs during swallowing analyzed by the motion analysis unit using the particle method, and analyzes the muscle activity rate based on the evaluation function.
[0008] Furthermore, the swallowing simulation method according to the present invention includes a storage step of storing a dynamic three-dimensional head and neck particle model in which a plurality of head and neck organs are modeled by a plurality of muscle particles having muscle fiber directions specified for each muscle type within a three-dimensional image, and a contraction stress based on the muscle fiber direction during swallowing is imparted to the muscle particles by a muscle activity rate; a muscle activity rate changing step of changing the value of the muscle activity rate set for each muscle type; a motion analysis step of analyzing the movement of the head and neck organs during swallowing in the three-dimensional image based on a particle method using the dynamic three-dimensional head and neck particle model in which the value of the muscle activity rate has been changed; and a muscle activity rate analysis step of calculating an evaluation function for evaluating the muscle activity rate based on the analysis result of the movement of the head and neck organs during swallowing analyzed by the particle method in the motion analysis step, and analyzing the muscle activity rate based on the evaluation function. Effect of the Invention
[0009] According to the present invention, a dynamic three-dimensional head and neck particle model in which muscle activity rates are set for muscle particles for each muscle type is used to analyze the movement of the head and neck organs during swallowing based on the particle method while changing the value of the muscle activity rate, and an evaluation function for evaluating the muscle activity rate is calculated based on the analysis result. As a result, in the present invention, the value of the muscle activity rate can be determined based on the evaluation function calculated each time the value of the muscle activity rate is changed, and a swallowing simulation device and a swallowing simulation method can be realized that can reproduce the movement of the head and neck organs during swallowing more accurately than ever before. [Brief description of the drawings]
[0010] [Figure 1] FIG. 2 is a block diagram showing a circuit configuration of the swallowing simulation device. [Diagram 2] FIG. 1 is a schematic diagram showing the construction of a dynamic three-dimensional head and neck particle model. [Diagram 3] FIG. 3 is a cross-sectional view showing the cross-sectional configuration of the dynamic three-dimensional head and neck particle model shown in FIG. 2 in the median plane. [Figure 4] FIG. 1 is a schematic diagram showing the state change of a dynamic three-dimensional head and neck particle model during swallowing. [Diagram 5] FIG. 1 is a schematic diagram for explaining a muscle particle. [Figure 6] FIG. 1 is a schematic diagram for explaining the muscle fiber direction of a muscle particle. [Figure 7] This is a graph showing the time changes in muscle activity rate during swallowing in the superior pharyngeal constrictor muscle glossopharyngeal portion, the orthopharyngeal constrictor muscle cornu minor pharynx upper portion, the orthopharyngeal constrictor muscle cornu minor pharynx lower portion, the orthopharyngeal constrictor muscle cornu major pharynx upper portion, the orthopharyngeal constrictor muscle cornu major pharynx lower portion, the inferior pharyngeal constrictor muscle upper thyropharyngeal portion, the inferior thyropharyngeal constrictor muscle lower portion, and the inferior pharyngeal constrictor muscle cricopharyngeal portion. [Figure 8] FIG. 13 is a schematic diagram for explaining contact forces acting on muscle particles in a dynamic three-dimensional particle-based head and neck model. [Figure 9] 11 is a flowchart showing a calculation process procedure when performing a swallowing simulation with a dynamic three-dimensional head and neck particle model using the Hamiltonian particle method (Hamiltonian MPS method). [Figure 10] 13 is a flowchart showing a calculation process procedure according to another embodiment. [Figure 11] FIG. 1 is a schematic diagram showing the construction of a simplified dynamic 3D head and neck particle model. [Figure 12] FIG. 12 is a schematic diagram showing the state change of the dynamic three-dimensional head and neck particle model shown in FIG. 11 during swallowing. [Figure 13] 13 is a flowchart showing a muscle activity rate analysis process according to a first method. [Figure 14] FIG. 13 is a schematic diagram showing the dynamic three-dimensional head and neck particle model when an arbitrary value of muscle activity rate αt is manually set for each muscle type. [Figure 15] 13 is a flowchart showing a manual setting process procedure. [Figure 16] 11 is a table for explaining muscle activity rates set by a manual setting process. [Figure 17A] 1 is a flowchart showing a static optimization process procedure (1). [Figure 17B] 13 is a flowchart showing a static optimization process procedure (2). [Figure 18]13 is a table showing an example of muscle activation rates αmt finally obtained by the static optimization processing procedure. [Figure 19] 13 is a flowchart showing a muscle activity rate analysis process according to a second method. [Figure 20] 13 is a flowchart showing a dynamic optimization process procedure. [Figure 21] FIG. 11 is a schematic diagram for explaining an optimization target time. [Figure 22A] 1 is a flowchart showing an analysis process procedure (1). [Figure 22B] 13 is a flowchart showing an analysis process procedure (2). [Figure 23] FIG. 2 is a schematic diagram for explaining one calculation process according to the particle method performed during the optimization target time. [Figure 24] 13 is a table showing an example of muscle activation rates αmt finally obtained by a dynamic optimization processing procedure in another embodiment in which relaxation processing is performed. [Diagram 25] 13 is a flowchart illustrating a dynamic optimization process according to another embodiment. [Figure 26] FIG. 13 is a schematic diagram for explaining an optimization target time in another embodiment. [Figure 27] 13 is a flowchart showing an analysis process procedure in which a relaxation process is performed. [Figure 28A] 13 is a graph showing changes over time in muscle activity rate when the relaxation degree β is 1. [Figure 28B] 13 is a graph showing the change over time in muscle activity rate when the relaxation degree β is 10. [Figure 28C] 13 is a graph showing the change over time in muscle activity rate when the relaxation degree β is 100. [Figure 28D] 13 is a graph showing the change over time in muscle activity rate when the relaxation degree β is 1000. [Figure 28E] 13 is a graph showing the change over time in muscle activity rate when the relaxation degree β is 10,000. [Figure 29A] 13 is a graph showing the trajectory of the hyoid bone when the relaxation degree β is 1. [Figure 29B]13 is a graph showing the trajectory of the hyoid bone when the relaxation degree β is 10. [Figure 29C] 13 is a graph showing the trajectory of the hyoid bone when the relaxation degree β is 100. [Figure 29D] 13 is a graph showing the trajectory of the hyoid bone when the relaxation degree β is 1000. [Figure 29E] 13 is a graph showing the trajectory of the hyoid bone when the relaxation degree β is 10,000. [Diagram 30] 11 is a table showing an example of the number of optimization calculations and an evaluation function. [Diagram 31] 1 is a table showing the relationship between muscle type and muscle activity rate. [Diagram 32] FIG. 13 is a schematic diagram showing a simplified dynamic three-dimensional head and neck particle model used in the verification test, the configuration of which is a dynamic three-dimensional head and neck particle model in which different muscle activity rates are set for the left and right stylohyoid muscles, and the configuration of which is a dynamic three-dimensional head and neck particle model in which the same muscle activity rate is set for the left and right stylohyoid muscles. [Diagram 33] 13 is a graph showing the coordinates of the center of gravity of the hyoid bone during swallowing in a dynamic three-dimensional head and neck particle model in which optimal muscle activity rates are set by static optimization processing. [Diagram 34] 13 is a graph showing the shift in the center of gravity of the hyoid bone during swallowing and the displacement from the reference position in a dynamic three-dimensional head and neck particle model in which optimal muscle activation rates are set by static optimization processing. [Diagram 35] 13 is a graph showing an evaluation function of the hyoid bone during swallowing and its displacement from a reference position in a dynamic three-dimensional head and neck particle model in which optimal muscle activity rates are set by static optimization processing. [Diagram 36] This is a graph showing the muscle activity rates of each muscle type during swallowing, optimized by static optimization processing, and the displacement of the hyoid bone during swallowing, in a dynamic three-dimensional head and neck particle model in which separate muscle activity rates are set for the left and right stylohyoid muscles. [Figure 37] This is a graph showing the muscle activity rate of each muscle type during swallowing, optimized by static optimization processing, and the displacement of the hyoid bone during swallowing, in a dynamic three-dimensional head and neck particle model in which the same muscle activity rate is set for the left and right stylohyoid muscles. [Figure 38] FIG. 1 is a schematic diagram showing the configuration of a dynamic three-dimensional head and neck particle model used in a validation test. [Figure 39A] FIG. 1 is a schematic diagram showing the dynamic three-dimensional head and neck particle model obtained by a verification test at the start of optimization calculations. [Figure 39B] FIG. 1 is a schematic diagram showing the dynamic three-dimensional head and neck particle model obtained by a verification test at the end of optimization calculations. [Diagram 40] 13 is a table showing an evaluation function calculated by a verification test. [Diagram 41] This is a schematic diagram showing the swallowing motion (1) of the dynamic 3D head and neck particle model obtained in the validation test, an image showing only the optimized muscle types, and an image showing an enlarged view of the hyoid bone with the muscle particles hidden. [Diagram 42] This is a schematic diagram showing the swallowing movement (2) of the overall image of the dynamic 3D head and neck particle model obtained in the verification test, an image showing only the optimized muscle types, and an image showing an enlarged view of the hyoid bone with the muscle particles hidden. [Figure 43A] 13 is a graph showing the trajectory of the center of gravity of the hyoid bone during a swallowing simulation when wu-=0.0. [Figure 43B] 13 is a graph showing the trajectory of the center of gravity of the hyoid bone during a swallowing simulation when wu-=0.1. [Figure 43C] 13 is a graph showing the trajectory of the center of gravity of the hyoid bone during swallowing simulation when wu-=0.5. [Diagram 44] 13 is a graph showing the trajectory of the center of gravity of the hyoid bone during each swallowing simulation when γ=1.0, 0.75, and 0.5. [Figure 45A] 13 is a graph showing the change over time in muscle activity rate α of the stylohyoid muscle (StyH), geniohyoid muscle (GH), and sternohyoid muscle (SteH) during a swallowing simulation when γ=1.0. [Figure 45B] 13 is a graph showing the change over time in muscle activity rate α of the stylohyoid muscle (StyH), geniohyoid muscle (GH), and sternohyoid muscle (SteH) during a swallowing simulation when γ=0.75. [Figure 45C] 13 is a graph showing the change over time in muscle activity rate α of the stylohyoid muscle (StyH), geniohyoid muscle (GH), and sternohyoid muscle (SteH) during a swallowing simulation when γ=0.5. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0011] Hereinafter, an embodiment of the present invention will be described in detail.
[0012] (1) <Overview of the present invention> 1 is a block diagram showing the overall configuration of a swallowing simulation device 1 of the present invention. The swallowing simulation device 1 includes a personal computer (also referred to as a PC) 2, an input unit 81, a display unit 4, and a storage unit 83. The input unit 81 is an input device such as a mouse or a keyboard, and outputs an operation command from a developer to the personal computer 2, and causes the personal computer 2 to execute various arithmetic processing according to the operation command. The storage unit 83 stores a dynamic three-dimensional head and neck particle model of particle display formed by the personal computer 2 (described later in FIG. 2), a dynamic three-dimensional head and neck particle model of particle display with simplified head and neck organs (described later in FIG. 11), setting conditions, analysis results, an evaluation function described later, muscle activity rate, etc.
[0013] The personal computer 2 forms, for example, a dynamic three-dimensional head and neck particle model (described later in FIG. 2) consisting of head and neck organs using a three-dimensional image, and models the orally ingested item as a pseudo-orally ingested item within the three-dimensional image. The personal computer 2 can analyze the movement of each head and neck organ in the dynamic three-dimensional head and neck particle model, and the behavior of the pseudo-orally ingested item during swallowing, within the three-dimensional image, using a particle method.
[0014] In addition, the personal computer 2 can also analyze only the movement of each head and neck organ during swallowing in the dynamic three-dimensional head and neck particle model within the three-dimensional image using the particle method, without placing a simulated orally ingested item within the three-dimensional image.
[0015] Such a dynamic three-dimensional head and neck particle model is formed by imitating, for example, the head and neck of a person with a swallowing disorder who is prone to aspiration, or the head and neck of a healthy person who is unlikely to aspiration. For example, a dynamic three-dimensional head and neck particle model modeled on the head and neck of a person with a swallowing disorder who is likely to aspiration can analyze the movement of each head and neck organ when a person with a swallowing disorder swallows, and the behavior of a simulated oral intake product when swallowing. On the other hand, a dynamic three-dimensional head and neck particle model modeled on the head and neck of a healthy person who is unlikely to aspiration can analyze the movement of each head and neck organ when a healthy person swallows, and the behavior of a simulated oral intake product when swallowing.
[0016] The analysis results obtained by the personal computer 2 are output to the display unit 4 and displayed on the display screen of the display unit 4. The display unit 4 is, for example, a display, and displays the three-dimensional image of the dynamic three-dimensional head and neck particle model output from the personal computer 2, the simulated orally ingested product, the analysis results, the calculated evaluation function, the muscle activity rate for each muscle type, etc. on the display screen. As a result, the display unit 4 allows the developer to visually confirm the movement of each head and neck organ in the dynamic three-dimensional head and neck particle model, the behavior of the simulated orally ingested product during swallowing, the analysis results, the evaluation function, the muscle activity rate for each muscle type, etc. In addition, in this embodiment, the three-dimensional image refers to a moving image in which the dynamic three-dimensional head and neck particle model is expressed in a virtual three-dimensional space, and frame images that constitute the moving image.
[0017] In the swallowing simulation device 1, a swallowing simulation can be performed using a dynamic three-dimensional head and neck particle model by changing the physical properties of the simulated orally ingested food item, such as the food bolus amount, viscosity, and specific gravity, and the dynamic three-dimensional head and neck particle model can also be used to confirm the presence or absence of aspiration. Furthermore, even if a simulated orally ingested food item is not set in the three-dimensional image, the swallowing simulation device 1 can perform a swallowing simulation using the dynamic three-dimensional head and neck particle model, and can analyze the movement of each head and neck organ of the dynamic three-dimensional head and neck particle model during swallowing.
[0018] In this embodiment, a particle method is used as an analysis method capable of expressing deformation of a liquid surface, splashes, etc., in which the liquid or solid to be analyzed is treated as particles, and the particle method is used to represent the movement (operation) of the head and neck organs in a dynamic three-dimensional head and neck particle model and the behavior of an orally ingested item in a three-dimensional image to perform a swallowing simulation. As the particle method, it is particularly preferable to apply the MPS (Moving Particle Semi-implicit) method (Koshizuka et al., Comput. Fluid Dynamics J, 4, 29-46, 1995). As the particle method when analyzing the behavior of a simulated orally ingested item during swallowing by a swallowing simulation, it is preferable to apply the MPS method or the Hamiltonian particle method (HMPS method).
[0019] In addition, it is preferable to apply the Hamiltonian particle method (Hamiltonian MPS method) as a particle method when analyzing the movement of each particle in the dynamic three-dimensional head and neck particle model 10c by swallowing simulation. In this embodiment, the case where the Hamiltonian particle method (Hamiltonian MPS method) is applied as a particle method for analyzing the movement of each particle in the dynamic three-dimensional head and neck particle model 10c by swallowing simulation will be described below.
[0020] In the particle method of this embodiment, not only the simulated orally ingested product is replaced with a particle, but also the head and neck organs in the dynamic three-dimensional head and neck particle model are replaced with a particle, and the physical quantity is calculated for each particle. As a result, it is possible to analyze the head and neck organs in the dynamic three-dimensional head and neck particle model and subtle changes during swallowing of the simulated orally ingested product. In this embodiment, the calculation process by the particle method performed using such a dynamic three-dimensional head and neck particle model is also called swallowing simulation.
[0021] Here, in the swallowing simulation device 1 of this embodiment, not only are the head and neck organs in the dynamic three-dimensional head and neck particle model replaced with particles, but further, for the particles forming the head and neck organs such as the oral cavity, pharynx, and larynx, the muscle fiber direction is specified within the three-dimensional image for each muscle type of the head and neck organ based on medical knowledge, and a contraction stress based on the muscle fiber direction is applied.
[0022] In this embodiment, the muscle fiber direction is specified in the three-dimensional image for each muscle type of the head and neck organs, and the particle to which the contraction stress based on the muscle fiber direction is applied is called a muscle particle. In the swallowing simulation device 1, the behavior of the muscle particles set for each muscle type during swallowing (temporal change in the contraction stress applied to the muscle fiber direction) is reproduced by adjusting the temporal change in the value of the muscle activity rate (described later).
[0023] Here, the muscle activity rate indicates the temporal change in the contraction stress of the muscle particles during swallowing, which is set for each muscle type with respect to the muscle particles to which a contraction stress based on the muscle fiber direction is applied during swallowing. In the swallowing simulation device 1, the value of the muscle activity rate that changes over time during swallowing can be changed for each muscle type, thereby changing the behavior of the muscle particles during swallowing for each muscle type. Therefore, by adjusting the value of the muscle activity rate to an optimal value for each muscle type, it is possible to realize the ideal movement of each head and neck organ during swallowing.
[0024] The swallowing simulation device 1 performs a swallowing simulation (computational processing using the particle method) using a dynamic three-dimensional head and neck particle model in which a predetermined value of muscle activity rate is set for each muscle type, and obtains analysis results regarding the behavior of muscle particles during swallowing (position and speed of muscle particles). The swallowing simulation device 1 stores target values (target position and target speed of muscle particles) indicating ideal muscle particle behavior (position and speed of muscle particles) during swallowing for each muscle type in advance in the storage unit 83, and calculates an evaluation function (described later) that evaluates the value of the muscle activity rate set during the swallowing simulation based on the analysis results obtained by the swallowing simulation and the target values.
[0025] The evaluation function is an index for judging whether the behavior of muscle particles that move based on the muscle activity rate coincides with the ideal behavior of muscle particles during swallowing. In the swallowing simulation device 1 of this embodiment, the evaluation function is calculated using the result of one cycle of calculation processing by the particle method (described later in FIG. 9 and FIG. 10) performed during the swallowing simulation using the dynamic three-dimensional head and neck particle model as the analysis result of the swallowing simulation. Then, in the swallowing simulation device 1, the value of the muscle activity rate is changed to an optimal value for each muscle type based on the analysis result of the evaluation function, and the swallowing simulation is performed again using the dynamic three-dimensional head and neck particle model with the changed muscle activity rate.
[0026] In this way, the swallowing simulation device 1 repeatedly changes the value of the muscle activity rate, performs one cycle of calculation processing using the particle method using a dynamic three-dimensional head and neck particle model with the changed value of the muscle activity rate, and calculates an evaluation function using the results of one cycle of calculation processing using the particle method.By comparing the multiple evaluation functions obtained, it is possible to ultimately determine the optimal value of the muscle activity rate.
[0027] In this way, the swallowing simulation device 1 can realize ideal movement of each head and neck organ during swallowing by appropriately adjusting the value of the muscle activity rate based on the analysis results of the evaluation function, and can reproduce the movement of the head and neck organs during swallowing and the behavior of the simulated oral ingested item more accurately than ever before.
[0028] (2) <Circuit configuration of the personal computer in the swallowing simulation device> Next, the personal computer 2 of the swallowing simulation device 1 will be described below. As shown in Fig. 1, the personal computer 2 includes a head and neck modeling unit 10, an organ motion setting unit 30, an orally ingested product physical property setting unit 40, a motion analysis unit 50, a physical property identification unit 70, and a control unit 90, as well as a muscle activity rate changing unit 80 and a muscle activity rate analysis unit 91. The head and neck modeling unit 10 forms, for example, (i) a dynamic three-dimensional head and neck particle model 10c that faithfully reproduces the head and neck organs as shown in Figs. 2 and 3, (ii) a dynamic three-dimensional head and neck particle model 50a that simplifies the head and neck organs as shown in Fig. 11, and a dynamic three-dimensional head and neck particle model 10g as shown in Fig. 38, using three-dimensional images.
[0029] In addition, FIG. 38, which will be described later, shows a dynamic three-dimensional head and neck particle model 10g that faithfully reproduces the head and neck organs, similar to FIG. 2 and FIG. 3, and was produced for a verification test. The dynamic three-dimensional head and neck particle model 10c in FIG. 2 and FIG. 3 and the dynamic three-dimensional head and neck particle model 10g in FIG. 38 are both models that faithfully reproduce each head and neck organ. Here, the dynamic three-dimensional head and neck particle model that faithfully reproduces the head and neck organs will be described below using the dynamic three-dimensional head and neck particle model 10c in FIG. 2 and FIG. 3.
[0030] Here, we will first explain the dynamic three-dimensional particle model 10c of the head and neck, which faithfully reproduces the head and neck organs as shown in Figures 2 and 3, and then explain the dynamic three-dimensional particle model 50a of the head and neck, which has simplified the head and neck organs as shown in Figure 11.
[0031] In addition, the following description will be given using a particle-displayed dynamic three-dimensional head and neck particle model 10c as shown in Figures 2 and 3, but the present invention is not limited to this. For example, a dynamic three-dimensional head and neck model may be applied that is created from the particle-displayed dynamic three-dimensional head and neck particle model 10c using a marching cubes method or the like, and that simply displays the surface of the head and neck organs without displaying each particle while leaving the setting information of each particle of the dynamic three-dimensional head and neck particle model 10c intact.
[0032] When the simulated orally ingested product formed within the three-dimensional image is displayed on the display unit 4, each particle that forms the simulated orally ingested product may not be displayed, and only the surface shape of the simulated orally ingested product generated by the marching cubes method or the like may be displayed.
[0033] The organ motion setting unit 30 sets the motion of each head and neck organ in the dynamic three-dimensional head and neck particle model 10c. The organ motion setting unit 30 in this embodiment includes a forced motion setting unit 31 and a muscle contraction motion setting unit 32. The forced motion setting unit 31 sets particles that are forced to move in the head and neck organs when the simulated orally ingested product is swallowed by the dynamic three-dimensional head and neck particle model 10c as forced movement particles (described later), and sets the motion of these multiple forced movement particles. The muscle contraction motion setting unit 32 specifies the muscle fiber direction in the three-dimensional image for each muscle type of the head and neck organ based on medical knowledge, sets particles to which a contraction stress based on the muscle fiber direction is applied as muscle particles, and sets the motion of the muscle particles based on the contraction stress when the simulated orally ingested product is swallowed.
[0034] As a result, the particles forming the dynamic three-dimensional head and neck particle model 10c are defined as one of three types: forced movement particles, muscle particles, and particles other than these forced movement particles and muscle particles. The dynamic three-dimensional head and neck particle model 10c can execute a swallowing simulation in which each head and neck organ moves, based on the setting state by the organ movement setting unit 30. In addition, when there is no particular need to distinguish between the forced movement particles, muscle particles, and particles other than these forced movement particles and muscle particles, the three types of particles, the forced movement particles, muscle particles, and particles other than these forced movement particles and muscle particles, will hereinafter be collectively referred to simply as particles.
[0035] The orally ingested product physical property setting unit 40 sets the physical property values of the orally ingested product, such as food and drink, medicine, or quasi-drug, as the analysis target, and forms a pseudo-orally ingested product that models the orally ingested product in a three-dimensional image. The orally ingested product physical property setting unit 40 can set multiple pseudo-orally ingested products, which are liquids, semi-solids, or solids, with different physical properties as the analysis target. Semi-solids include, for example, jellies, and solids include, for example, tablets.
[0036] In the present embodiment, the orally ingested product physical property setting unit 40 sets, as physical property values of the orally ingested product, for example, the density [g / mL] of the food bolus to be the orally ingested product, the amount of the food bolus to be swallowed by the dynamic three-dimensional head and neck particle model 10c [mL], the surface tension [N / m], the contact angle at each head and neck organ, and the slip coefficient at each head and neck organ. Note that the slip coefficient here is a parameter that controls the wettability and water repellency of the surface of the living body and the surface of the food bolus (orally ingested product), and can be considered as the apparent viscosity at the contact surface. When the slip coefficient is large, the friction at the interface increases, which has the effect of braking the movement of the food bolus. When the slip coefficient is small, the friction at the interface decreases, and when the slip coefficient is 0, the surface becomes like a mirror surface. A slip coefficient of 1 means that a friction effect equivalent to the viscosity of the fluid is given to the interface. The slip coefficient is determined by analyzing the wettability, water repellency, etc. of the assumed orally ingested product.
[0037] The orally ingested product property setting unit 40 sets the contact angles at the pharynx, larynx, tongue, and soft palate in the dynamic three-dimensional head and neck particle model 10c as the contact angles at each head and neck organ. The orally ingested product property setting unit 40 also sets the slip coefficients at the pharynx, larynx, tongue, and soft palate in the dynamic three-dimensional head and neck particle model 10c as the slip coefficients at each head and neck organ.
[0038] In this embodiment, not only the above-mentioned physical property values are used as the physical property values of the orally ingested product, but also, for example, when the orally ingested product is liquid, physical property values such as liquid volume, viscosity, surface tension, specific gravity, thermal conductivity, and specific heat may be set. When the orally ingested product is solid, physical property values such as shape, size, elastic modulus, tensile strength, yield point, yield stress, shear rate dependency of viscosity, dynamic viscoelasticity, static viscoelasticity, compressive stress, breaking stress, breaking strain, hardness, adhesiveness, cohesiveness, thermal conductivity, and specific heat may be set. Furthermore, when the orally ingested product is semi-solid (has plasticity but no fluidity), physical property values such as amount, viscosity, specific gravity, yield point, yield point stress, shear rate dependency of viscosity, dynamic viscoelasticity, static viscoelasticity, compressive stress, adhesiveness, and cohesiveness may be set.
[0039] The movement analysis unit 50 performs a swallowing simulation using a particle method to analyze the movement of the head and neck organs when swallowing in the dynamic three-dimensional head and neck particle model 10c, and the behavior of the simulated orally ingested item when swallowed in conjunction with the movement of the head and neck organs.
[0040] Here, Fig. 3 is a cross-sectional view showing the cross-sectional configuration in the median plane of the dynamic three-dimensional particle model of the head and neck 10c shown in Fig. 2. In the dynamic three-dimensional particle model of the head and neck 10c shown in Fig. 2 and Fig. 3, the progressive wave-like wave motion of the tongue 12, the rotational motion of the epiglottis 15a, the reciprocating motion of the larynx 15, the muscle contraction motion of the pharyngeal region 14, and other movements are reproduced by analysis using the particle method, and move a simulated orally ingested item (not shown) inserted inside the head and neck. The movement of the simulated orally ingested item is also analyzed using the particle method. The simulated orally ingested item is treated as particles whether it is solid, semi-solid, or liquid.
[0041] The motion analysis unit 50 performs a swallowing simulation using a particle method, for example, after the orally ingested product physical property setting unit 40 changes the physical property values of the simulated orally ingested product, or the muscle activity rate changing unit 80 changes the value of the muscle activity rate. When the physical property values of the simulated orally ingested product are changed, the motion analysis unit 50 can obtain an analysis result showing a change in the path taken when the simulated orally ingested product is swallowed by the swallowing simulation using the particle method.
[0042] Furthermore, when the muscle activity rate value is changed for each muscle type by the muscle activity rate changing unit 80, the motion analysis unit 50 can obtain analysis results showing changes in the behavior of the traveling wave wave motion of the tongue 12, the lifting motion of the soft palate 13b, the inversion motion of the epiglottis 15a, the lifting motion of the larynx 15, the adduction motion of the vocal cords 15c, the forward movement of the arytenoids 15b, and the contraction and lifting motion of the pharynx 14, by simulating swallowing using a particle method.
[0043] The physical property identifying unit 70 estimates the food bolus amount, viscosity, and shear rate of the simulated orally ingested product that can avoid aspiration, based on the analysis results of the motion analysis unit 50. Among these, the food bolus amount and viscosity of the simulated orally ingested product are physical property values set by the orally ingested product physical property setting unit 40.
[0044] The control unit 90 controls each unit of the personal computer 2 to execute various functions of the swallowing simulation device 1. The control unit 90 has a swallowing simulator (analysis software) in its built-in memory.
[0045] The muscle activity rate changing unit 80 changes the muscle activity rate value set for each muscle type in the dynamic three-dimensional head and neck particle model 10c in accordance with an operation command from the input unit 81. This allows the motion analysis unit 50 to perform a swallowing simulation using the particle method with the dynamic three-dimensional head and neck particle model 10c in which the muscle activity rate value has been changed for each muscle type.
[0046] The muscle activity rate analysis unit 91 includes an evaluation function calculation unit 92, a steady state determination unit 93, a gradient calculation unit 94, a search direction determination unit 95, a line search processing unit 96, an optimization number setting unit 97, and a relaxation processing unit 98. The evaluation function calculation unit 92 calculates an evaluation function for evaluating the muscle activity rate value set for each muscle type during the swallowing simulation, using the analysis result obtained by the swallowing simulation of the dynamic three-dimensional head and neck particle model 10c by the motion analysis unit 50 and the muscle activity rate value set at that time.
[0047] The steady state determination unit 93 determines whether the multiple evaluation functions calculated for each swallowing simulation have reached a steady state in which the values of the evaluation functions remain almost unchanged, based on the analysis results of the swallowing simulation in which the muscle activity rate values were changed for each muscle type.
[0048] The gradient calculation unit 94, the search direction determination unit 95, and the line search processing unit 96 perform calculations to search for an optimal muscle activity rate using the evaluation function and the muscle activity rate. The optimization number setting unit 97 sets the number of optimization calculations of the evaluation function calculated by the muscle activity rate analysis unit 91.
[0049] The details of the evaluation function calculation unit 92, steady state determination unit 93, gradient calculation unit 94, search direction determination unit 95, line search processing unit 96, optimization number setting unit 97, and relaxation processing unit 98, which constitute the muscle activity rate analysis unit 91, will be described later. Note that in this embodiment, relaxation processing (described in other embodiments described later) is not performed, so the relaxation processing unit 98 will not be described, but will be described in other embodiments in which relaxation processing is performed.
[0050] (3)<Construction of a dynamic 3D particle model of the head and neck that faithfully reproduces the head and neck organs> Before describing the muscle activity rate changing unit 80 and the muscle activity rate analyzing unit 91, which are characteristic components of this embodiment, the dynamic three-dimensional head and neck particle model 10c formed by the head and neck modeling unit 10 will be described below. In this embodiment, there are two types of dynamic three-dimensional head and neck particle model: the dynamic three-dimensional head and neck particle model 10c that faithfully reproduces the head and neck organs, and the dynamic three-dimensional head and neck particle model 50a that simplifies the head and neck organs as shown in Fig. 11. Here, the dynamic three-dimensional head and neck particle model 10c that faithfully reproduces the head and neck organs will be described first, and then the dynamic three-dimensional head and neck particle model 50a that simplifies the head and neck organs will be described.
[0051] As described above, the display unit 4 displays, for example, a dynamic three-dimensional head and neck particle model 10c and a simulated orally ingested item (not shown) as shown in FIG. 2, and as a result of the analysis by the movement analysis unit 50, the movement of the head and neck organs when the simulated orally ingested item is swallowed in the dynamic three-dimensional head and neck particle model 10c and the behavior of the simulated orally ingested item are displayed as moving images.
[0052] In this embodiment, a particle-display dynamic three-dimensional head and neck particle model 10c and a simulated orally ingested product are used to present the movement of the head and neck organs and the behavior of the simulated orally ingested product through moving images. However, the present invention is not limited to this, and the movement of the head and neck organs and the behavior of the simulated orally ingested product may be presented through moving images using a surface-display dynamic three-dimensional head and neck model and a simulated orally ingested product in which only the surface shape is displayed without displaying each individual particle.
[0053] 2 and 3, the X-axis indicates the left-right direction of the body (the surface normal direction of the median plane) perpendicular to the median plane of the dynamic three-dimensional head and neck particle model 10c, the Y-axis indicates the front-back direction of the body perpendicular to the X-axis and parallel to the median plane of the dynamic three-dimensional head and neck particle model 10c, and the Z-axis indicates the up-down direction of the body perpendicular to the X-axis and Y-axis.
[0054] As shown in Figures 2 and 3, the dynamic three-dimensional head and neck particle model 10c has, as head and neck organs, a tongue 12 including the geniohyoid muscle, a larynx 15, vocal cords 15c, arytenoids 15b, epiglottis 15a, trachea 16, pharyngeal region 14 (including pharyngeal wall 18a and pharyngeal mucosa 18b), palate 13 (including hard palate 13a and soft palate 13b), and esophagus 17 (including esophageal inlet 17a and esophageal wall 17b). In this embodiment, the tongue 12, larynx 15, vocal cords 15c, arytenoids 15b, epiglottis 15a, trachea 16, pharyngeal region 14, palate 13, esophagus 17, etc. are collectively referred to as head and neck organs, but each of these is also simply referred to as a head and neck organ. Note that the pseudo-oral intake product is not shown in Figures 2 and 3.
[0055] In this embodiment, a simulated orally ingested item (not shown) is represented by particles, and the head and neck organs in the dynamic three-dimensional head and neck particle model 10c (in this embodiment, the tongue 12, larynx 15, vocal cords 15c, arytenoids 15b, epiglottis 15a, trachea 16, pharynx 14, palate 13, and esophagus 17) are represented by particles.
[0056] However, as described above, when the analysis result of the swallowing simulation of the simulated orally ingested product by the dynamic three-dimensional head and neck particle model 10c is visually confirmed by the developer, etc., using moving images, it is preferable to display the surface shapes of the head and neck organs of the dynamic three-dimensional head and neck particle model 10c and the simulated orally ingested product. This simplifies the display form of the head and neck organs and the simulated orally ingested product compared to the case where each particle forming the head and neck organs and the simulated orally ingested product is displayed, and the developer, etc., can easily confirm the movement of the head and neck organs and the behavior of the simulated orally ingested product.
[0057] (4)<Creation of a dynamic 3D particle model of the head and neck> (4-1) <Creating a static three-dimensional head and neck model> Next, a method for creating a dynamic three-dimensional particle model 10c of the head and neck represented by particles will be described below. First, the positions of the pharynx 14 and the entrance to the esophagus 17a are estimated based on the structure of the head and neck as understood by medical knowledge and the shapes of the palate 13, tongue 12, and trachea 16 that can be roughly read from CT (Computed Tomography) images. The structures of the tongue 12, palate 13, pharynx 14, epiglottis 15a, larynx 15, and esophagus 17 are modeled using CG (Computer Graphics) software (Autodesk 3ds Max, etc.), and a static initial shape model (not shown) that represents the head and neck organs involved in swallowing in three dimensions (three-dimensional structure) is created.
[0058] The static initial shape model thus obtained is overlaid with contrast images (front and side views) taken during swallowing in a videofluoroscopic examination of swallowing (VF) to correct the three-dimensional structure, and a static three-dimensional head and neck model (not shown) is created in which the three-dimensional shape of the head and neck organs of the subject at rest is depicted by computer graphics. Alternatively, a static three-dimensional head and neck model can be created based on four-dimensional computed tomography (CT) images (4DCT images) taken during swallowing. Such a static three-dimensional head and neck model is created by the head and neck modeling unit 10.
[0059] (4-2) <Particle modeling of static 3D head and neck model> Next, based on this static three-dimensional head and neck model, a dynamic three-dimensional head and neck particle model 10c is produced as shown in Figures 2 and 3. A method for producing the dynamic three-dimensional head and neck particle model 10c will be described below.
[0060] In this case, the surfaces of the head and neck organs (tongue 12, palate 13, pharynx 14, epiglottis 15a, larynx 15, and esophagus 17 in Figures 2 and 3) in a static three-dimensional head and neck model (not shown) created using CG are identified, and particles are placed within each region of each head and neck organ, with the surfaces of the head and neck organs as boundaries, as shown in Figures 2 and 3.
[0061] The particles in this embodiment are three-dimensional spherical particles (simply referred to as particles in this embodiment) having a three-dimensional shape in a three-dimensional image. For example, when creating a life-size model of the head and neck of an average-sized baby or adult male using a three-dimensional image, it is desirable for the particle diameter to be approximately 0.1 mm to 3.0 mm, and more preferably, the diameter should be approximately 0.6 mm to 1.5 mm.
[0062] In addition, in the static three-dimensional head and neck model created in the three-dimensional image, it is desirable to form at least two particles in the direction of the epiglottis thickness (for example, about 3.0 mm in an adult and about 1.5 mm in an infant). If the diameter of the particles is too small, the computational load on the personal computer 2 becomes too large, which is undesirable, whereas if the diameter of the particles is too large, the fine movements of the head and neck organs cannot be reproduced, so it is desirable for the diameter of the particles to be within the above range.
[0063] In this embodiment, a case is described in which three-dimensional spherical particles are used as particles for forming the head and neck organs. However, the present invention is not limited to this, and the head and neck organs may be formed using particles of various other shapes, such as rectangular prism-shaped particles and polygonal-shaped particles.
[0064] When creating the dynamic three-dimensional head and neck particle model 10c, the surfaces of each head and neck organ (tongue 12, palate 13, pharynx 14, epiglottis 15a, larynx 15, and esophagus 17) of a static three-dimensional head and neck model (not shown) created using CG based on CT images and VF images are first identified, and then the particles are placed without any gaps within the area surrounded by the surfaces so that the particles are in contact with each other, thereby modeling each head and neck organ using particles.
[0065] In addition, when performing a swallowing simulation in which a simulated oral intake item is swallowed using the dynamic three-dimensional head and neck particle model 10c, the surface of the simulated oral intake item created using CG is also identified, and then the particles are placed without any gaps so that the particles are in contact with each other within the area surrounded by the surface, thereby modeling the simulated oral intake item using particles.
[0066] In this manner, the process of creating each head and neck organ and simulated orally ingested item of the static three-dimensional head and neck model using particles is performed by the head and neck modeling unit 10.
[0067] Here, in the dynamic three-dimensional head and neck particle model 10c in this embodiment, among the multiple particles, particles in a predetermined region are set as forced movement particles that are forced to move in the head and neck organs when the simulated oral intake product is swallowed. Also, in the dynamic three-dimensional head and neck particle model 10c in this embodiment, among the multiple particles, particles in a predetermined region other than the particles that are the forced movement particles are set as muscle particles to which a contraction stress based on the muscle fiber direction is applied.
[0068] In addition, particles other than the forced movement particles and muscle particles are particles whose positions to which they are forced to move in the head and neck organs during swallowing (i.e., coordinates to which they move in the three-dimensional image during swallowing) are not specified, and whose contraction stress in the direction of muscle fibers, unlike muscle particles, is not specified. When performing a swallowing simulation using the dynamic three-dimensional head and neck particle model 10c, the moving positions, etc. of such particles other than the forced movement particles and muscle particles are analyzed using a conventional particle method.
[0069] Details of conventional swallowing simulations of particles other than forced-movement particles and muscle particles are disclosed in the literature "Kikuchi, T., Michiwaki, Y., Koshizuka, S., Kamiya, T., and Toyama Y., "Numerical simulation of interaction between organs and food bolus during swallowing and aspiration," Computers in Biology and Medicine, 80, (2017), pp. 114-123." Therefore, we will omit the explanation here and focus on forced-movement particles and muscle particles in the following explanation.
[0070] Regarding the forced moving particles described below, a simulation using a particle method is also disclosed in the literature "Kikuchi, T., Michiwaki, Y., Kamiya, T. et al. Comp. Part. Mech. (2015) 2: 247. "Human swallowing simulation based on videofluorography images using Hamiltonian MPS method".
[0071] (4-3)<Setting of forced moving particles in dynamic 3D head and neck particle model> The dynamic three-dimensional particle model of the head and neck 10c has forced movement particles and muscle particles among the particles forming the head and neck organs. First, the setting of the forced movement particles in the dynamic three-dimensional particle model of the head and neck will be described.
[0072] In the dynamic three-dimensional particle model of the head and neck 10c, some particles of the head and neck organs are set as forced movement particles, and the movement of the muscles considered to be the main active muscles during swallowing is modeled by using these forced movement particles. In this case, in the dynamic three-dimensional particle model of the head and neck 10c, particles that give a rigid forced displacement during swallowing are selected from the particles forming the tongue 12, palate 13, pharynx 14, epiglottis 15a, larynx 15, esophagus 17, etc., and set as forced movement particles 19.
[0073] The forced movement particles 19 are selected based on anatomical knowledge and knowledge from medical image analysis research so as to reflect the muscle movement of the head and neck organs when a subject actually swallows an orally ingested product. In this embodiment, the head and neck organs are traced in VF images and 4DCT images obtained when a subject swallows a specific orally ingested product, and the forced movement particles 19 to be forced to move within the dynamic three-dimensional head and neck particle model 10c are selected.
[0074] In addition, based on VF images and 4DCT images obtained when the subject swallows a specified orally ingested item, the position to which each forcedly moved particle 19 moves within the three-dimensional image is determined at predetermined time intervals (e.g., 0.1 S) from the start of swallowing to the end of swallowing, and the time and position at the time of swallowing is set for each forcedly moved particle 19.
[0075] That is, in the dynamic three-dimensional head and neck particle model 10c, for example, a forced moving particle 19 whose coordinates on the X-axis, Y-axis, and Z-axis in the three-dimensional image are (0.0, 0.2, -0.2) at 0.0S, which is the start of swallowing, is set to move to coordinates (0.0, 0.2, 0.0) at 0.1S from the start of swallowing, and to coordinates (0.0, 0.3, 0.3) at 0.2S.
[0076] 2 and 3 indicate forced movement particles 19 in this embodiment, which are set, for example, in parts of tongue 12, palate 13, larynx 15, etc. In tongue 12 of this embodiment, island-like particle groups 19a each consisting of a plurality of forced movement particles 19 are set at predetermined intervals along the tongue surface, as shown in Fig. 3. In palate 13, island-like particle groups 19b each consisting of a plurality of forced movement particles 19 are set at a location on the oral cavity side of the soft palate.
[0077] In the larynx 15 of this embodiment, an island-like particle group 19c where a plurality of forced movement particles 19 are gathered is set near the epiglottic vallecula, an island-like particle group 19d where a plurality of forced movement particles 19 are gathered is set near the laryngeal prominence, and a particle group 19e where a plurality of forced movement particles 19 are gathered is set near the posterior cricoarytenoid muscle. Furthermore, in the esophagus 17, island-like particle groups 19f where a plurality of forced movement particles 19 are gathered are set at predetermined intervals along the tube wall on the airway side.
[0078] Here, Figure 4 is a schematic diagram showing the state changes of dynamic three-dimensional head and neck particle models 10c1, 10c2, 10c3, and 10c4 when forced-movement particles 19 are moved along movement trajectory lines 22a, 22b, 22c, and 22d (hereinafter collectively referred to as movement trajectory lines 22) in dynamic three-dimensional head and neck particle model 10c shown in Figure 3, where part of the trajectory of the forced-movement particles 19 when swallowing is represented by movement trajectory lines 22a, 22b, 22c, and 22d.
[0079] For example, 22a indicates the movement trajectory line of the forced-movement particles 19 set on the tongue 12, 22b indicates the movement trajectory line of the forced-movement particles 19 set on the soft palate of the palate 13, 22c indicates the movement trajectory line of the forced-movement particles 19 set on the larynx 15, and 22d indicates the movement trajectory line of the forced-movement particles 19 set on the tubal wall of the esophagus 17.
[0080] Figure 4 shows an example of a dynamic three-dimensional head and neck particle model 10c that swallows a specified simulated oral intake item in approximately 12 S, with dynamic three-dimensional head and neck particle model 10c1 at 0 S, which is the start of swallowing, dynamic three-dimensional head and neck particle model 10c2 approximately 7 S after the start of swallowing, dynamic three-dimensional head and neck particle model 10c3 approximately 9 S after the start of swallowing, and dynamic three-dimensional head and neck particle model 10c4 approximately 11 S after the start of swallowing.
[0081] In this way, in the dynamic three-dimensional head and neck particle model 10c, the forced moving particles 19 are provided at predetermined positions of the head and neck organs, and the positions and times to which each forced moving particle 19 moves during swallowing are set in advance, thereby reproducing the basic movements of the head and neck organs during swallowing (traveling wave wave movement, rotational movement, up and down movement, back and forth movement, contraction movement, etc.). The movements of the forced moving particles 19 in such dynamic three-dimensional head and neck particle model 10c are set by the forced movement setting unit 31 of the organ movement setting unit 30.
[0082] (4-4)<Setting of muscle particles in a dynamic 3D head and neck particle model> In order to accurately reproduce the behavior of the pharyngeal part 14 and the like during swallowing in the dynamic three-dimensional head and neck particle model 10c, it is desirable to reproduce the movement in which the length of the wall of the pharyngeal part 14 and the like is shortened during swallowing. Therefore, in the dynamic three-dimensional head and neck particle model 10c in this embodiment, not only is a rigid forced displacement given to the particles of the pharyngeal part 14 and the like, but a muscle fiber direction is set for each muscle particle in the three-dimensional image for each muscle type of the pharyngeal part 14 and the like, and an optimal contraction stress based on the muscle fiber direction is given to the muscle particle by the muscle activity rate, thereby accurately reproducing the behavior of the dynamic three-dimensional head and neck particle model 10c during swallowing. Note that the particles that give the contraction stress based on the muscle fiber direction in the dynamic three-dimensional head and neck particle model 10c are called muscle particles.
[0083] In the dynamic three-dimensional head and neck particle model 10c, for example, when muscle particles are set from among the particles 20a forming the tongue 12, muscle body regions ER1, ER2, etc. where the tongue 12 expands and contracts during swallowing are identified in the three-dimensional image based on anatomical knowledge, VF images, 4DCT images, etc., as shown in Fig. 5, and particles 20a present in each muscle body region ER1, ER2 are searched for. For example, the particle 20a in the muscle body region ER1 is set as the muscle particle of the tongue 12, and the muscle fiber direction is defined for each muscle particle in the virtual space of the three-dimensional image.
[0084] The details of the muscle fiber direction set for the muscle particle will be described later. Based on anatomical knowledge, VF images, 4DCT images, etc., a plurality of lines A are set in the space of the muscle body region ER1 of the tongue 12 in the direction in which muscle contraction occurs during swallowing, and the weighted average of the directions of the adjacent lines A for each muscle particle is set as the muscle fiber direction. In this embodiment, for example, two lines are specified, a first line segment closest to the muscle particle and a second line segment second closest to the muscle particle, and the muscle fiber direction is obtained by averaging the directions of the first line segment and the second line segment with weights according to the closeness of the distance from the muscle particle to the first line segment and the distance from the muscle particle to the second line segment. However, the method of obtaining the muscle fiber direction does not have to be this method. For example, a muscle fiber direction that is more smoothly distributed in space can be obtained by performing radial basis function interpolation using all the lines defined in the muscle body region ER1.
[0085] In addition, in the dynamic three-dimensional head and neck particle model 10c, when the simulated orally ingested product is swallowed, the time change in the contraction stress of the muscle particles that occurs for each muscle type of the contractile muscle of the head and neck organs is set as the muscle activity rate, and the magnitude of the contraction stress during swallowing is set by the muscle activity rate. Note that the swallowing simulation device 1 of this embodiment can calculate an evaluation function by a muscle activity rate analysis process described later, and determine the value of this muscle activity rate based on the evaluation function.
[0086] Here, first, focusing on the pharyngeal region 14 in the dynamic three-dimensional head and neck particle model 10c, the muscle fiber directions set for each muscle type in the pharyngeal region 14 will be described below. The left diagram of Fig. 6 is a schematic diagram of a muscle fiber model 10d showing directions Aa, Ab, Ac, Ad, Ae, Af, Ag, and Ah in which the constrictor muscles of the pharyngeal region 14 run in the static three-dimensional head and neck model 10b. The right diagram of Fig. 6 is a schematic diagram of a dynamic three-dimensional head and neck particle model 10e showing the muscle fiber directions of each muscle particle in the pharyngeal region 14 in the dynamic three-dimensional head and neck particle model 10c based on directions Aa, Ab, Ac, Ad, Ae, Af, Ag, and Ah in which the pharyngeal constrictor muscles run shown in the muscle fiber model 10d in the left diagram.
[0087] In this embodiment, as shown in the muscle fiber model 10d and the dynamic three-dimensional head and neck particle model 10e, based on anatomical knowledge, the pharyngeal portion 14 is divided into the glossopharyngeal portion of the superior pharyngeal constrictor muscle 14a, the cornu lesser pharyngeal upper portion of the oropharyngeal constrictor muscle 14b, the cornu lesser pharyngeal lower portion of the oropharyngeal constrictor muscle 14c, the cornu greater pharyngeal upper portion of the oropharyngeal constrictor muscle 14d, the cornu greater pharyngeal lower portion of the oropharyngeal constrictor muscle 14e, the upper thyropharyngeal portion of the inferior pharyngeal constrictor muscle 14f, the lower thyropharyngeal portion of the inferior pharyngeal constrictor muscle 14g, and the cricopharyngeal portion of the inferior pharyngeal constrictor muscle 14h.
[0088] For the superior pharyngeal constrictor muscle glossopharyngeal portion 14a, the orthopharyngeal constrictor muscle cornu lesser pharynx upper portion 14b, the orthopharyngeal constrictor muscle cornu lesser pharynx lower portion 14c, the orthopharyngeal constrictor muscle cornu greater pharynx upper portion 14d, the orthopharyngeal constrictor muscle cornu greater pharynx lower portion 14e, the inferior pharyngeal constrictor muscle thyropharyngeal upper portion 14f, the inferior pharyngeal constrictor muscle thyropharyngeal lower portion 14g, and the inferior pharyngeal constrictor muscle cricopharyngeal portion 14h, particles that become muscle particles are arranged without any gaps within each region, and modeling is performed using particles.
[0089] The directions Aa, Ab, Ac, Ad, Ae, Af, Ag, and Ah of the superior pharyngeal constrictor muscle (glossopharyngeal portion) 14a, the orthopharyngeal constrictor muscle (angle lesser pharyngeal portion) 14b, the orthopharyngeal constrictor muscle (angle lesser pharyngeal portion) 14c, the orthopharyngeal constrictor muscle (angle greater pharyngeal portion) 14d, the orthopharyngeal constrictor muscle (angle greater pharyngeal portion) 14e, the inferior pharyngeal constrictor muscle (thyropharyngeal portion) 14f, the inferior pharyngeal constrictor muscle (thyropharyngeal portion) 14g, and the inferior pharyngeal constrictor muscle (cricopharyngeal portion) 14h are determined by setting the fine muscle contraction directions that occur at each part of each constrictor muscle during swallowing as line segments based on anatomical knowledge and VF images, and then calculating the weighted average of the directions of each line segment in the vicinity of each muscle particle. For the sake of convenience, the directions Aa, Ab, Ac, Ad, Ae, Af, Ag, and Ah of the contractile muscles shown on the left side of Fig. 6 are approximate directions of the muscle fibers. Each muscle particle in the pharyngeal region 14 defines the muscle fiber direction according to the directions Aa, Ab, Ac, Ad, Ae, Af, Ag, and Ah of the contractile muscles.
[0090] For each of the muscle particles provided in the above-mentioned superior pharyngeal constrictor muscle glossopharyngeal portion 14a, the middle pharyngeal constrictor muscle cornu minor pharyngeal upper portion 14b, the middle pharyngeal constrictor muscle cornu minor pharyngeal lower portion 14c, the middle pharyngeal constrictor muscle cornu major pharyngeal upper portion 14d, the middle pharyngeal constrictor muscle cornu major pharyngeal lower portion 14e, the hypopharyngeal constrictor muscle thyropharyngeal upper portion 14f, the hypopharyngeal constrictor muscle thyropharyngeal lower portion 14g, and the hypopharyngeal constrictor muscle cricopharyngeal portion 14h, a muscle activity rate α is set in accordance with the passage of time during swallowing, as shown in FIG. 7, for example.
[0091] Here, Fig. 7 shows an example of the muscle activity rate α finally specified by the muscle activity rate analysis process in the swallowing simulation device 1 of this embodiment. The swallowing simulation device 1 of this embodiment is characterized in that the value of the muscle activity rate α set for each muscle type as shown in Fig. 7 can be set to an optimal value that can accurately reproduce the movement of the head and neck organs during swallowing by performing muscle activity rate analysis process. The swallowing simulation device 1 of this embodiment repeatedly performs swallowing simulation while changing the value of the muscle activity rate α in the dynamic three-dimensional head and neck particle model 10c, calculates an evaluation function every time the swallowing simulation is performed, and determines the optimal muscle activity rate α by analyzing the evaluation function calculated each time.
[0092] (5)<Structural analysis method for dynamic 3D head and neck particle model> Before explaining the muscle activity rate analysis process, the structural analysis method of the dynamic three-dimensional head and neck particle model performed during the muscle activity rate process will be explained below. There are two types of structural analysis methods for the dynamic three-dimensional head and neck particle model: (i) a structural analysis method for the dynamic three-dimensional head and neck particle model incorporating internal and external forces, and (ii) a structural analysis method for the dynamic three-dimensional head and neck particle model incorporating only internal forces. The following explains each method in order.
[0093] The internal force is the elastic force f applied to the muscle particle by the particle method. i,elastic and the artificial potential force f applied to the muscle particle. i,artificial and the viscous force acting on the muscle particle, f i,viscous The external force is the contact force f applied to a muscle particle when it comes into contact with another muscle particle. i,contact and the fluid force acting on the muscle particle from the simulated ingested product, f i,interaction He says.
[0094] (5-1)<Structural analysis method for dynamic 3D head and neck particle model incorporating internal and external forces> (5-1-1) <Governing equations when incorporating internal and external forces> In this case, the swallowing simulation device 1 can perform simulation analysis using three-dimensional images based on the particle method to determine the movement of each head and neck organ when a simulated oral intake item is swallowed using the dynamic three-dimensional head and neck particle model 10c, and the behavior of the simulated oral intake item when swallowed.
[0095] In this embodiment, when performing a swallowing simulation by displaying the motion of the head and neck organs in the dynamic three-dimensional head and neck particle model 10c and the behavior of a simulated orally ingested item in a three-dimensional image, the muscle particles of the head and neck organs in the dynamic three-dimensional head and neck particle model 10c are analyzed as Mooney-Rivlin bodies by a particle method (for example, Hamiltonian particle method: Hamiltonian MPS method) in the motion analysis unit 50. Note that, here, the following description will focus mainly on muscle particles among the particles forming the dynamic three-dimensional head and neck particle model 10c.
[0096] In this case, the governing equation that determines the motion of each muscle particle in the dynamic three-dimensional head and neck particle model 10c is given by the following equation (1).
number
[0097] ρ is the density of muscle particle i, v i is the velocity vector of muscle particle i, the subscript i is an identifier for identifying muscle particle i, t is time, and ∂v i / ∂t is the acceleration of muscle particle i at the current time.
[0098] f i,elastic is the elastic force applied to muscle particle i by the particle method. f i,artificial is the artificial potential force applied to muscle particle i. f i,viscous is the viscous force acting on muscle particle i. f i,contact is the contact force applied to muscle particle i when it comes into contact with another muscle particle. f i,interactionis the fluid force from the simulated orally ingested product acting on the muscle particle i. Below, these will be explained in order. During the swallowing simulation, the position and speed of each muscle particle in the dynamic three-dimensional head and neck particle model 10c in the three-dimensional image are calculated from the above formula (1), and the muscle particles move based on the obtained position and speed. In other words, during the swallowing simulation, each muscle particle does not simply move in the direction of the muscle fiber, but moves in the three-dimensional image according to the position and speed calculated based on the above formula (1).
[0099] <Elastic force f acting on muscle particle i i,elastic > In the Hamiltonian MPS method, the relative position vector between muscle particle i and another muscle particle j at the current time is expressed as r ij Let r be the relative position vector between muscle particle i and muscle particle j at the initial time 0. 0 i,j Then, the relative position vector is r ij , r 0 i,j Using the deformation gradient tensor F i can be calculated by the following formula (2). Note that the initial time 0 indicates 0S, which is the start time of the swallowing simulation.
number
[0100] The symbol with a cross in the circle above indicates a tensor product, and w 0 ij is the weight function between muscle particle i and muscle particle j at the initial time 0. A i is expressed by the following equation (3).
number
[0101] In addition, the elastic force f acting on muscle particle i i,elastic can be obtained by differentiating the sum of elastic strain potential energies W=Π:F, and is expressed by the following equation (4).
number
[0102] That is, the elastic force f acting on muscle particle i i,elastic is expressed by the following equation (5).
number
[0103] In the above formula (4), j is the first Piola-Kirchhoff stress tensor in muscle particle j. F j is the deformation gradient tensor of muscle particle j. A j is the above equation (3) for muscle particle j.
[0104] S i is the second Piola-Kirchhoff stress tensor for muscle particle i, which represents the stress applied to muscle particle i. Note that S j is the second Piola-Kirchhoff stress tensor in muscle particle j, where the second Piola-Kirchhoff stress tensor S is expressed by the following equation (6).
number
[0105] S passive represents the passive stress due to the movement of other muscle particles and forced moving particles, and S active represents the active contractile stress, which is the active contractile stress based on the muscle fiber orientation.
[0106] S i,active is the active contractile stress S in muscle particle i active and is expressed by the following equation (7).
number
[0107] S i,m is expressed by the following arithmetic expression (8). The subscript i is an identifier for identifying the muscle particle i, and the subscript m is an identifier defined for each muscle type to identify the muscle type of the constrictor muscle of the head and neck organs. In this embodiment, the muscle type of the constrictor muscle of the head and neck organs is the muscle type in the pharyngeal part 14, and the subscript m indicates any one of the superior pharyngeal constrictor muscle glossopharyngeal part 14a, the middle pharyngeal constrictor muscle cornu minor pharyngeal upper part 14b, the middle pharyngeal constrictor muscle cornu minor pharyngeal lower part 14c, the middle pharyngeal constrictor muscle cornu major pharyngeal upper part 14d, the middle pharyngeal constrictor muscle cornu major pharyngeal lower part 14e, the inferior pharyngeal constrictor muscle thyropharyngeal upper part 14f, the inferior pharyngeal constrictor muscle thyropharyngeal lower part 14g, and the inferior pharyngeal constrictor muscle cricopharyngeal part 14h.
number
[0108] α m indicates the muscle activity rate that indicates the change over time in the contractile stress of muscle particles in muscle type m when a simulated orally ingested product is swallowed using the dynamic three-dimensional head and neck particle model 10c. In this embodiment, the muscle activity rate α m is set.
[0109] In this embodiment, the muscle activity rate is indicated by a symbol such as α, α m , α t , α m t , α k,0 However, these muscle activity rates α and α m , muscle activity rate α t , muscle activity rate α m t , muscle activity rate α k,0 The numerical values shown as ", etc." indicate the same thing.
[0110] f max is the muscle activity rate α m is the maximum active contractile stress when is at its maximum value, and f lis the correction coefficient for the active contraction stress based on the muscle fiber length at the current time. 0 i,m indicates the muscle fiber direction at the initial time 0 set for muscle particle i of muscle type m.
[0111] In this embodiment, the muscle activity rate α that can be changed by the muscle activity rate changing unit 80 is m is 0≦α m ≦1, and the maximum active contraction stress f max is α m When f = 1 max = 700kPa. Correction coefficient f l is expressed by the following equation (9).
number
[0112] In the above formula (9), I4 is expressed by the following formula (10), and I4 0 is a constant that indicates the optimal length state. i is the right Cauchy-Green deformation tensor of muscle particle i.
number
[0113] Next, the passive stress S passive The passive stress S passive is expressed by the following equation (11).
number
[0114] W is expressed by the following equation (12). C is the right Cauchy-Green deformation tensor, C=F T F is the deformation gradient tensor, and F T denotes the transpose of F.
number
[0115] In the above equation (12), C1, C2, and D1 are the material constants of the Mooney-Rivlin body, and I - 1 is the first reduction invariant of C, and J is J=det F.
[0116] <Artificial potential force f applied to muscle particle i i,artificial > Artificial potential force f of muscle particle i i,artificial is a function that suppresses the singular displacement mode and the vibration caused by it, which is determined by the Hamiltonian MPS method, and works to cancel the error of the deformation gradient tensor F. i,artificial is expressed by the following equation (13). i art is expressed by the following equation (14).
number
number
[0117] In addition, this artificial potential force f i,artificialThe details of the calculation method are disclosed in the literature "Kikuchi, T., Michiwaki, Y., Koshizuka, S., Kamiya, T., and Toyama Y., "Numerical simulation of interaction between organs and food bolus during swallowing and aspiration," Computers in Biology and Medicine, 80, (2017), pp. 114-123." and the literature "Kikuchi, T., Michiwaki, Y., Kamiya, T. et al. Comp. Part. Mech. (2015) 2: 247. "Human swallowing simulation based on videofluorography images using Hamiltonian MPS method", so the explanation will be omitted here.
[0118] <Viscosity force f acting on muscle particle i i,viscous > Viscous force f acting on muscle particle i i,viscous damps the velocity of muscle particle i and stabilizes the calculation, and is expressed by the following equation (15).
number
[0119] ρ is the density of muscle particle i. ν ela is the viscosity of the elastic body. d is the dimensionality. λ and n 0 is a constant of the Hamiltonian particle method (Hamiltonian MPS method).
[0120] This viscous force f i,viscousThe details of the calculation method are disclosed in the same paper as above, “Kikuchi, T., Michiwaki, Y., Koshizuka, S., Kamiya, T., and Toyama Y., “Numerical simulation of interaction between organs and food bolus during swallowing and aspiration,” Computers in Biology and Medicine, 80, (2017), pp. 114-123.” Therefore, the explanation will be omitted here.
[0121] <Contact force f applied to muscle particle i i,contact > Contact force f applied to muscle particle i i,contact is the contact force between muscle particles, and the normal force f i,nor and friction force f i,tan Here, a contact force is applied to muscle particle i that is embedded in a wall (a wall determined from other muscle particles j that are in contact with it) by the penalty method. In the penalty method, which is applied as a contact boundary condition with other objects, a penalty force is applied to muscle particle i that is in contact with other objects, i.e., a normal force f i,nor is given to muscle particle i.
number
[0122] k is the penalty coefficient, p is the amount of penetration, and n is the normal force f i,nor As shown in Figure 8, the normal force f i,nor applies a force proportional to the amount of penetration p into the wall in the direction of the wall's normal vector n.
[0123] Also, the friction force f i,tan is expressed by the following equation (17).
number
[0124] ρ i is the density of muscle particle i, and μ tan is the friction coefficient, and v iw,tan is the component of the relative velocity between muscle particle i and the wall that is perpendicular to the normal vector n of the wall.
[0125] Such a contact force f i,contact The details of the calculation method are disclosed in the literature "Takahiro Kikuchi, Yukihiro Michiwaki, Seiichi Koshizuka, Satoshi Kamiya, Takashi Nagata, Nobuko Kamino, Yoshio Toyama. Uniaxial compression simulation of a hyperelastic body model using the Hamiltonian MPS method with the penalty method introduced as the wall boundary condition. Transactions of the Japan Society for Computational Engineering and Science, 2014:20140010, 2014," so the explanation will be omitted here.
[0126] <Fluid force f acting on muscle particle i from a simulated ingested product i,interaction > Here, we performed a fluid analysis with the structural particles (muscle particles) as wall particles, and calculated the reaction force of the force exerted by the structural particles on the fluid particles (particles of the pseudo-oral ingested product) as the fluid force f i,interaction The governing equation for an incompressible fluid can be expressed by the following equation (18). Here, the pseudo-oral ingestible product to be swallowed is analyzed as a Newtonian fluid using the E-MPS method (Explicit MPS method).
[0127] Fluid force f acting on muscle particle i from a simulated ingestible product i,interaction The governing equation is given by the following equation (18).
number
[0128] Here, the first term on the right-hand side of the above equation (18) is a pressure gradient term, the second term on the right-hand side is a viscosity term, the third term on the right-hand side is a gravity term, and the fourth term on the right-hand side is a surface tension term.
[0129] v is the velocity of the fluid (simulated oral intake), ρ is the density of the fluid, P is the pressure of the fluid, ν is the dynamic viscosity of the fluid, g is the gravitational acceleration, f surface tension is the surface tension. Each term on the right hand side of the above equation (18) except for the gravitational acceleration is formulated in a form that includes the effect of the wall boundary condition that the fluid particle receives from the wall particle.
[0130] The details of the calculation method for the governing equations of such incompressible fluid are disclosed in the literature "Masatoshi Daichi, Seiichi Koshizuka, Mikio Sakai. Development of MPS Explicit Algorithm for Free Surface Flow Analysis. Transactions of the Japan Society for Computational Engineering and Science, 2010:20100013, 2010.", "Yukito Suzuki. Research on High Precision Particle Method and Multiphysics Simulator. Doctoral Thesis, 2007.", and "Masahiro Kondo, Seiichi Koshizuka, Masato Takimoto. Surface Tension Model Using Inter-Particle Potential Force in MPS Method. Transactions of the Japan Society for Computational Engineering and Science, 2007:20070021, 2007.", so the explanation will be omitted here.
[0131] Regarding the fluid pressure P in the above equation (18), the pressure P acting on the muscle particle i from the fluid is i is expressed by the following equation (19).
number
[0132] c is the speed of sound in the fluid, n i is the sum of weight functions between muscle particle i and all other muscle particles and fluid particles, and is called the particle number density of muscle particle i. n0 is the particle number density of muscle particle i at the initial time 0. ▽P i is expressed by the following equation (20). P j is the pressure exerted by the fluid on muscle particle j.
number
[0133] surface tension f surface tension is expressed by the following equation (21).
number
[0134] C ST ij is the coefficient of potential force. φ is expressed by (22) below.
number
[0135] l0 is the initial nearest neighbor distance, and r e is the radius of influence of the potential force.
[0136] (5-1-2) <Calculation procedure using the Hamiltonian particle method (Hamiltonian MPS method) when incorporating internal and external forces> In the swallowing simulation device 1 of this embodiment, each equation described in the above "(5-1-1) <Governing equation when internal and external forces are incorporated>" is stored in the motion analysis unit 50. In the swallowing simulation device 1, the physical property values of the pseudo orally ingested product are set by the orally ingested product physical property setting unit 40, and further, the muscle activity rate changing unit 80 changes the muscle activity rate α m Once these are set, the movement of the muscle particles of the dynamic three-dimensional head and neck particle model 10c (the position r and speed v at which the muscle particles move) and the behavior of the simulated orally ingested item can be calculated based on each equation stored in the motion analysis unit 50.
[0137] Based on such calculation results, the swallowing simulation device 1 can present, as moving images, the dynamic three-dimensional head and neck particle model 10c and a swallowing simulation in which the dynamic three-dimensional head and neck particle model 10c swallows a simulated orally ingested item.
[0138] In this embodiment, the movement of muscle particles of the dynamic three-dimensional head and neck particle model 10c (the position r where the muscle particles move) is calculated based on each equation. t ´ and speed v t Once the muscle activity rate αm The developer or the like who operates the swallowing simulation device 1 can calculate an evaluation function for evaluating the muscle activity rate α m The value of and the movement of the muscle particles of the dynamic three-dimensional head and neck particle model 10c based on each equation (the position r where the muscle particles move) t ´ and speed v t ´) and calculate the evaluation function, and by analyzing the evaluation function, the muscle activity rate α m The value of is specified.
[0139] First, based on the above-mentioned equations, the movement of the muscle particles of the dynamic three-dimensional head and neck particle model 10c (the position r where the muscle particles move) is calculated. t ´ and speed v t The calculation process for calculating the suction pressure (P) during the swallowing simulation will be described below.
[0140] Here, Fig. 9 is a flowchart showing the calculation procedure when performing a swallowing simulation with the dynamic three-dimensional head and neck particle model 10c using the Hamiltonian particle method (Hamiltonian MPS method). Note that, here, the following description focuses on a specific muscle particle i of the dynamic three-dimensional head and neck particle model 10c, but the following calculation process is performed for each muscle particle of the dynamic three-dimensional head and neck particle model 10c.
[0141] In addition, since the positions (coordinates within the three-dimensional image) to which the forcedly moving particles set in the dynamic three-dimensional head and neck particle model 10c move from the start of swallowing to the end of swallowing are predetermined, the movement of the forcedly moving particles within the three-dimensional image during the swallowing simulation is not determined by the calculation processing described below.
[0142] However, the following calculation process is also performed on the forced moving particles. This is because the stress, etc. calculated for the forced moving particles affect muscle particles and other particles (particles other than the forced moving particles and muscle particles). That is, the second Piola-Kirchhoff stress tensor calculated for the forced moving particles affects the calculation of the elastic force (the above formula (5)) in step S3 when, for example, calculating muscle particle i. Also, the deformation gradient tensor calculated for the forced moving particles affects the calculation of the artificial potential force (the above formula (13)) in step S3 and the analysis of the contact force (the above formulas (16) and (17)) in step S4 when, for example, calculating muscle particle i.
[0143] In addition, the following calculation process is performed for particles other than the forced movement particles and muscle particles in the dynamic three-dimensional head and neck particle model 10c. In this case, the active contraction stress S active In other words, for particles other than the forced moving particles and muscle particles, the above-mentioned active contraction stress S active An operation that does not exist is performed.
[0144] In the case of a muscle particle i of the dynamic three-dimensional particle model of the head and neck 10c, first, in step S1, the motion analysis unit 50 calculates the position r of the muscle particle i of the dynamic three-dimensional particle model of the head and neck 10c at the time step (time) t. t and velocity v t Then, the process moves to the next step S2. In the initial step S1, the position r t =r 0 This is the initial shape of the dynamic three-dimensional head and neck particle model 10c itself, and the velocity v t =v 0 is 0. In addition, in the second and subsequent steps S1 after returning from step S10, the position r t ´ and speed v t ´, position r t and velocity v t Used as.
[0145] In step S2, the motion analysis unit 50 calculates the fluid force f from the simulated orally ingested product acting on the muscle particle i. i,interaction For this, the viscosity term in the second term on the right side of the above equation (18), the gravity term in the third term on the right side, and the surface tension term in the fourth term on the right side are obtained, and the process proceeds to the next step S3. Note that this motion analysis is a solution method in which the right side of the above equation (18) is not obtained simultaneously, but is obtained separately in steps S2 and S7, so the gradient pressure term in the first term on the right side of the above equation (18) is obtained in step S7.
[0146] In step S3, the motion analysis unit 50 performs structural analysis by calculating the elastic force f i,elastic is calculated based on the above equation (5), and the artificial potential force f i,artificial is calculated based on the above equation (13), and the viscous force f acting on muscle particle i is i,viscous is calculated based on the above formula (15), and the process proceeds to the next step S4.
[0147] In this embodiment, the elastic force f applied to the muscle particle i is i,elastic When calculating the passive stress S based on the above formula (6), passive and the active contraction stress S, which is the active contraction stress based on the muscle fiber direction. active This can be reflected in the muscle particles i, making it possible to reproduce the movement of the muscle particles i of the head and neck organs when swallowing a simulated oral intake product more accurately than ever before.
[0148] In step S4, the motion analysis unit 50 performs contact analysis by calculating the contact force f i,contact is calculated based on the above equations (16) and (17), and the process proceeds to the next step S5.
[0149] In step S5, the motion analysis unit 50 calculates the elastic force f i,elastic and the artificial potential force f applied to muscle particle i i,artificial and the viscous force f acting on muscle particle i i,viscous and the contact force f acting on muscle particle i i,contact Furthermore, the fluid force acting on muscle particle i from the simulated oral ingested product f i,interactionFrom the viscosity term in the second term on the right-hand side of equation (18), the gravity term in the third term on the right-hand side, and the surface tension term in the fourth term on the right-hand side of equation (18), the virtual acceleration ∂v i From the calculation result thus obtained, the tentative position r' and velocity v' of the muscle particle i are found, and the process proceeds to the next step S6.
[0150] In step S6, the motion analysis unit 50 calculates the pressure P from the above equation (19) using the tentative position r'. i In step S7, the motion analysis unit 50 calculates the pressure P i Using the above equation (20), the gradient pressure term in the first term on the right side of the above equation (18) is obtained, and the process proceeds to step S8.
[0151] As a result, the motion analysis unit 50 uses the gradient pressure term in the first term on the right side of the above equation (18) in step S7 to calculate the fluid force f i,interaction can be sought.
[0152] In step S8, the motion analysis unit 50 performs rigid body calculation by calculating the elastic force f i,elastic and the artificial potential force f applied to muscle particle i i,artificial and the viscous force f acting on muscle particle i i,viscous and the contact force f acting on muscle particle i i,contact and the fluid force acting on muscle particle i from the simulated ingested product, f i,interaction Using the above equation (1), the acceleration of muscle particle i, ∂v i / ∂t is calculated and the process proceeds to the next step S9.
[0153] In step S9, the motion analysis unit 50 calculates the acceleration ∂v i Based on / ∂t, the position r t ´ and speed v t The provisional position r' and velocity v' calculated in step S5 are then added to the position r' calculated in step S9. t ´ and speed v t´ and proceed to the next step S10.
[0154] In step S10, the motion analysis unit 50 advances the time step t to t=t+1 (t←t+1), returns to step S1, and performs the above-mentioned steps S1 to S9 to obtain the position r of the muscle particle i at the next time step (t+1). t+1 ´ and speed v t+1 In this manner, in the swallowing simulation device 1, the movement of each head and neck organ in the dynamic three-dimensional head and neck particle model 10c and the behavior related to the swallowing of the simulated orally ingested item can be analyzed in a three-dimensional image using the particle method, and the state of swallowing in the dynamic three-dimensional head and neck particle model 10c can be presented as a moving image on the display screen of the display unit 4.
[0155] When performing a swallowing simulation of the dynamic three-dimensional head and neck particle model 10c from the start of swallowing to the end of swallowing, the motion analysis unit 50 repeats the above-mentioned steps S1 to S10 at the time from the start of swallowing to the end of swallowing. Note that, among these, the analysis result of the swallowing simulation for calculating an evaluation function in the muscle activity rate analysis unit 91 described later is the result obtained in one cycle of calculation processing SRa of steps S1 to S9 (the position r calculated in step S9) among the results obtained in the calculation processing cycle in which steps S1 to S10 are repeated. t ´ and speed v t ´).
[0156] (5-2)<Structural analysis method for dynamic 3D particle model of head and neck incorporating only internal forces> (5-2-1) <Governing equations when only internal forces are included> Next, a structural analysis method of a dynamic three-dimensional head and neck particle model incorporating only internal forces will be described below. In this case, the swallowing simulation device 1 does not set a pseudo-oral ingested item in the three-dimensional image, and performs simulation analysis of the movement of each head and neck organ during swallowing in the dynamic three-dimensional head and neck particle model 10c in the three-dimensional image based on the particle method. Since the swallowing simulation device 1 does not set a pseudo-oral ingested item in the three-dimensional image, the behavior of the pseudo-oral ingested item cannot be analyzed visually, but the movement of each head and neck organ during swallowing can be analyzed relatively quickly with a small computational processing load.
[0157] In this embodiment as well, when performing a swallowing simulation by representing the movement of the head and neck organs in the dynamic three-dimensional head and neck particle model 10c in a three-dimensional image, the muscle particles of the head and neck organs in the dynamic three-dimensional head and neck particle model 10c are analyzed by the motion analysis unit 50 using a particle method (e.g., the Hamiltonian particle method: Hamiltonian MPS method) as a Mooney-Rivlin body.
[0158] In this case, the governing equation that determines the motion of each muscle particle in the dynamic three-dimensional head and neck particle model 10c is given by the following equation (23).
number
[0159] That is, when the movement of each head and neck organ during swallowing in the dynamic three-dimensional head and neck particle model 10c is simulated and analyzed in a three-dimensional image based on the particle method without setting a pseudo-oral ingested item in the three-dimensional image, the contact force f i,contact and fluid force f i,interaction Without providing a calculation term for the external force (see equation (1) above), the elastic force f i,elastic and artificial potential force f i,artificial and viscous force f i,viscous The motion of each muscle particle in the dynamic three-dimensional head and neck particle model 10c is determined only by the internal force calculation terms.
[0160] In addition, the elastic force f in the above equation (23) i,elasticand artificial potential force f i,artificial and viscous force f i,viscous These are the contents that were explained in the above "(5-1-1) <Governing equations when internal and external forces are incorporated>", so the explanation thereof will be omitted here to avoid duplication.
[0161] (5-2-2) <Calculation procedure using the Hamiltonian MPS method when only internal forces are incorporated> Next, the calculation process procedure by the Hamiltonian particle method (Hamiltonian MPS method) performed using the above formula (23) will be described below. In the swallowing simulation device 1 of this embodiment, each formula explained in the above "(5-2-1) <Governing equation when only internal force is incorporated>" is stored in the motion analysis unit 50.
[0162] The swallowing simulation device 1 changes the muscle activity rate α m When these are set, the motion of the muscle particles of the dynamic three-dimensional head and neck particle model 10c (the position r where the muscle particles move) is calculated based on each equation stored in the motion analysis unit 50. t ´ and speed v t Based on such a calculation result, the swallowing simulation device 1 can present the swallowing simulation of the dynamic three-dimensional head and neck particle model 10c as a moving image.
[0163] In this embodiment, the movement of the muscle particles of the dynamic three-dimensional head and neck particle model 10c (the position r where the muscle particles move) is calculated based on each equation. t ´ and speed v t Once these are calculated, the muscle activity rate α m In this embodiment, the value of the muscle activity rate α in the dynamic three-dimensional particle model of the head and neck 10c is changed, and the movement of the muscle particles of the dynamic three-dimensional particle model of the head and neck 10c based on each equation (the position r where the muscle particles move) is calculated. t ´ and speed v t´) and calculate the evaluation function, and by analyzing the evaluation function, the muscle activity rate α m The value of is specified.
[0164] FIG. 10, in which the same reference numerals are used for parts corresponding to those in FIG. 9, is a flowchart showing the computational procedure for performing a swallowing simulation using the dynamic three-dimensional head and neck particle model 10c by using the Hamiltonian particle method (Hamiltonian MPS method) using the above equation (23).
[0165] Note that the following explanation focuses on a specific muscle particle i of the dynamic three-dimensional head and neck particle model 10c. However, as the forced movement particles and other particles (particles other than the forced movement particles and muscle particles) are the same as those explained in the above "(5-1-2) <Calculation processing procedure using the Hamiltonian particle method (Hamiltonian MPS method) when internal and external forces are incorporated>", explanations thereof will be omitted.
[0166] In the case of a muscle particle i of the dynamic three-dimensional particle model of the head and neck 10c, first, in step S1, the motion analysis unit 50 calculates the position r of the muscle particle i of the dynamic three-dimensional particle model of the head and neck 10c at the time step (time) t. t and velocity v t Then, the process moves to the next step S3. In the initial step S1, the position r t =r 0 This is the initial shape of the dynamic three-dimensional head and neck particle model 10c itself, and the velocity v t =v 0 will be 0.
[0167] Next, in step S3, the motion analysis unit 50 performs structural analysis by calculating the elastic force f i,elastic is calculated based on the above equation (5), and the artificial potential force f i,artificial is calculated based on the above equation (13), and the viscous force f acting on muscle particle i is i,viscous is calculated based on the above formula (15), and the process proceeds to the next step S8.
[0168] In step S8, the motion analysis unit 50 performs rigid body calculation by calculating the elastic force f i,elastic and the artificial potential force f applied to muscle particle i i,artificial and the viscous force f acting on muscle particle i i,viscous Using the above equation (23), the acceleration of muscle particle i, ∂v i / ∂t is calculated and the process proceeds to the next step S9.
[0169] In step S9, the motion analysis unit 50 calculates the acceleration ∂v i Based on / ∂t, the position r t ´ and speed v t ´ and calculate the position r t and velocity v t and the position r calculated in step S9. t ´ and speed v t ´ and proceed to the next step S10.
[0170] In step S10, the motion analysis unit 50 advances the time step t to t+1 (t←t+1), returns to step S1, and performs the above-mentioned steps S1, S3, S8, and S9 to obtain the position r of the muscle particle i at the next time step (t+1). t+1 ´ and speed v t+1 In this manner, in the swallowing simulation device 1, the movement of each head and neck organ in the dynamic three-dimensional head and neck particle model 10c and the behavior related to the swallowing of the simulated orally ingested item can be analyzed in a three-dimensional image using the particle method, and the state of swallowing in the dynamic three-dimensional head and neck particle model 10c can be presented as a moving image on the display screen of the display unit 4.
[0171] When performing a swallowing simulation of the dynamic three-dimensional head and neck particle model 10c from the start of swallowing to the end of swallowing, the motion analysis unit 50 repeats the above-mentioned steps S1 to S10 from the start of swallowing to the end of swallowing. When using this "(5-2-2) <Calculation processing procedure by the Hamiltonian particle method (Hamiltonian MPS method) when only internal forces are incorporated>", the analysis result of the swallowing simulation used when calculating an evaluation function in the muscle activity rate analysis unit 91 described later is the result obtained in the calculation processing SRb of one cycle of steps S1, S3, and S8 to S10 (the position r calculated in step S9) out of the repeated cycles of steps S1, S3, and S8 to S10. t ´ and speed v t ´).
[0172] (6)<Dynamic 3D particle model of the head and neck with simplified head and neck organs> In the above-described embodiment, the dynamic three-dimensional head and neck particle model 10c that faithfully reproduces the shape and relative size of the head and neck organs has been described as a dynamic three-dimensional head and neck particle model for performing swallowing simulation. However, the present invention is not limited to this, and a dynamic three-dimensional head and neck particle model 50a in which the head and neck organs are simplified as shown in FIG. 11 may also be used.
[0173] In a simplified three-dimensional dynamic head and neck particle model 50a as shown in FIG. 11, for example, depiction of complex head and neck organs such as the tongue, palate, and larynx is omitted, and in order to analyze the behavior of the hyoid bone 52 during swallowing, the model is composed of skulls 51a, 51b, hyoid bone 52, mandible 53, sternum 54, stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58.
[0174] Furthermore, in the simplified three-dimensional dynamic head and neck particle model 50a, the skull 51a, 51b, hyoid bone 52, mandible 53, sternum 54, stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58 are not faithfully reproduced in terms of their respective shapes and relative sizes, but are formed by simple geometric figures such as simple rectangular prisms or linear bodies.
[0175] As a result, in the dynamic three-dimensional head and neck particle model 50a, it is possible to more easily confirm the behavior of the skull 51a, 51b, hyoid bone 52, and mandible 53 during swallowing, as well as the expansion and contraction movements of the stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58 during swallowing.
[0176] In this embodiment, in order to analyze the behavior of the hyoid bone 52 during swallowing, the skull 51a, 51b, the mandible 53, the sternum 54, the stylohyoid muscles 56a, 56b, the geniohyoid muscle 57, and the sternohyoid muscle 58 that affect the behavior of the hyoid bone 52 during swallowing are provided, but the present invention is not limited to this. For example, the behavior of other head and neck organs such as the epiglottis or thyroid cartilage during swallowing may be analyzed. In this case, by providing head and neck organs that affect the behavior of the epiglottis or thyroid cartilage, the behavior of the epiglottis or thyroid cartilage during swallowing can be analyzed by the behavior of simple geometric figures.
[0177] In the dynamic three-dimensional head and neck particle model 50a shown in FIG. 11, the sternum 54 is formed in a rectangular parallelepiped shape by a plurality of fixed particles in the three-dimensional image, and the sternum 54 is fixed at a predetermined lower position in the three-dimensional image. The skulls 51a and 51b are formed in a rectangular parallelepiped shape by a plurality of forced moving particles in the three-dimensional image, and are arranged side by side in the left-right direction (X-axis direction) of the dynamic three-dimensional head and neck particle model 50a at a predetermined upper position in the three-dimensional image. The mandible 53 is formed in a rectangular parallelepiped shape by a plurality of forced moving particles in the three-dimensional image, and is arranged between the sternum 54 and the skulls 51a and 51b in the three-dimensional image, and is arranged on the front side of the body in the front-back direction (Y-axis direction) of the dynamic three-dimensional head and neck particle model 50a.
[0178] The hyoid bone 52 is formed in a rectangular parallelepiped shape by a plurality of particles (particles other than forced movement particles and muscle particles) in the three-dimensional image, and is arranged so as to be located between the skulls 51a, 51b and the mandible 53, between the skulls 51a, 51b and the sternum 54, and between the mandible 53 and the sternum 54 in the three-dimensional image, and is arranged in an approximately central position of the dynamic three-dimensional head and neck particle model 50a at the start of swallowing. In addition, the hyoid bone 52 has a longitudinal direction in the left-right direction (X-axis direction) of the body of the dynamic three-dimensional head and neck particle model 50c, and the longitudinal length is selected to be shorter than the distance between the skulls 51a, 51b.
[0179] The stylohyoid muscles 56a, 56b are formed in a linear / rod shape by multiple muscle particles in the three-dimensional image, and the stylohyoid muscle 56a connects one end of the hyoid bone 52 to one skull 51a, and the stylohyoid muscle 56b connects the other end of the hyoid bone 52 to the other skull 51b.
[0180] The geniohyoid muscle 57 is formed in a linear / rod shape by a plurality of muscle particles in the three-dimensional image, and connects the hyoid bone 52 and the mandible 53. The sternohyoid muscle 58 is formed in a linear / rod shape by a plurality of muscle particles in the three-dimensional image, and connects the hyoid bone 52 and the sternum 54.
[0181] In order to accurately reproduce the behavior of the hyoid bone 52 and the like during swallowing in the dynamic three-dimensional head and neck particle model 50a, it is desirable to reproduce the movement in which the lengths of the stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58 are shortened during swallowing. Therefore, in the dynamic three-dimensional head and neck particle model 50a in this embodiment, not only is a rigid forced displacement given to the forced movement particles of the skull 51a, 51b and the mandible 53, but a muscle fiber direction is set for each muscle particle in the three-dimensional image for each muscle type of the stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58, and an optimal contraction stress based on the muscle fiber direction is given to the muscle particles by the muscle activity rate, thereby accurately reproducing the behavior of the hyoid bone 52 and the like during swallowing in the dynamic three-dimensional head and neck particle model 50a.
[0182] Such dynamic three-dimensional head and neck particle model 50a can be created by the following procedure. Dynamic three-dimensional head and neck particle model 50a determines the positions of rectangular parallelepiped skull 51a, 51b, hyoid bone 52, mandible 53, and sternum 54 in a three-dimensional image based on, for example, anatomical knowledge, VF images, 4DCT images, etc. Then, from the positions of skull 51a, 51b, hyoid bone 52, mandible 53, and sternum 54, the positions of linear / rod-shaped stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58 that expand and contract during swallowing are determined in the three-dimensional image, and particles present in the linear / rod-shaped regions of stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58 are searched for.
[0183] Particles within the regions of the stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58 are regarded as muscle particles, and muscle fiber directions are defined for the muscle particles for each muscle type in the virtual space of the three-dimensional image. As for the muscle fiber directions set for the muscle particles, similarly to the above-described embodiment, multiple directions in which muscle contraction occurs during swallowing are set as line segments within the space within the regions of the stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58 based on anatomical knowledge, VF images, 4DCT images, etc., and the weighted average of the directions of each nearby line segment is regarded as the muscle fiber direction for each muscle particle.
[0184] However, the method for determining the muscle fiber directions does not have to be this method. For example, radial basis function interpolation can be performed using all line segments defined within the areas of the stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58, to obtain muscle fiber directions that are more smoothly distributed in space.
[0185] Also, in the dynamic three-dimensional head and neck particle model 50a, the temporal change in contraction stress of muscle particles occurring for each muscle type of the stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58 is set as the muscle activity rate, and the magnitude of the contraction stress during swallowing is set based on the muscle activity rate. Note that the swallowing simulation device 1 of this embodiment can calculate an evaluation function by a muscle activity rate analysis process described later, and determine the value of this muscle activity rate based on the evaluation function.
[0186] FIG. 12 is a schematic diagram showing changes in the state of dynamic three-dimensional head and neck particle model 50a shown in FIG. 11 when swallowing behavior is set for the forced movement particles of skull 51a, 51b and mandible 53 and optimal muscle activity rates are set for the muscle particles of stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58.
[0187] FIG. 12 shows a dynamic three-dimensional head and neck particle model 50a1 approximately 0.70 S after the start of swallowing, a dynamic three-dimensional head and neck particle model 50a2 approximately 1.40 S after the start of swallowing, a dynamic three-dimensional head and neck particle model 50a3 approximately 1.90 S after the start of swallowing, and a dynamic three-dimensional head and neck particle model 50a4 approximately 2.37 S after the start of swallowing.
[0188] In this way, in the dynamic three-dimensional head and neck particle model 50a, by setting the behavior of the forced movement particles of the skull 51a, 51b and the mandible 53 during swallowing and optimally setting the muscle activity rate α for the muscle particles of the stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58, it is possible to reproduce the accurate behavior of the hyoid bone 52 and the like during swallowing.
[0189] (7) <Muscle activity rate analysis processing> Next, a muscle activity rate analysis process that uses an evaluation function to analyze the muscle activity rate, which is a characteristic feature of this embodiment, will be described below. Note that, here, the muscle activity rate analysis process when accurately reproducing the behavior of the hyoid bone 52 during swallowing in the simplified dynamic three-dimensional head and neck particle model 50a shown in Fig. 11 will be described as an example.
[0190] In the swallowing simulation device 1, after the above-mentioned dynamic three-dimensional head and neck particle model 50a is created within a three-dimensional image, in order to accurately reproduce the behavior of the hyoid bone 52 in the dynamic three-dimensional head and neck particle model 50a during swallowing, it is necessary to set the muscle activity rate of the muscle particles to an optimal value for each muscle type.
[0191] In the swallowing simulation device 1, in order to determine such muscle activity rate values, a predetermined value of muscle activity rate is set for each muscle type, and then one cycle of calculation processing SRb using the particle method during the swallowing simulation is performed using the dynamic three-dimensional head and neck particle model 50a in which the muscle activity rate is set, and the obtained calculation processing result is used as the analysis result of the swallowing simulation to calculate an evaluation function.
[0192] In addition, in the dynamic three-dimensional head and neck particle model 50c in which the head and neck organs are simplified, the path through each head and neck organ through which the pseudo-oral ingested product passes is also simplified. Therefore, when performing a swallowing simulation using the dynamic three-dimensional head and neck particle model 50c in which the head and neck organs are simplified, the pseudo-oral ingested product is not set in the three-dimensional image, and only the movement of each head and neck organ during swallowing is simulated and analyzed in the three-dimensional image based on the particle method. Therefore, in the dynamic three-dimensional head and neck particle model 50a, a swallowing simulation is performed according to the above-mentioned "(5-2) <Structural analysis method of dynamic three-dimensional head and neck particle model incorporating only internal forces>".
[0193] In the swallowing simulation device 1, an evaluation function is calculated based on the changed muscle activity rate value and the result of one cycle of calculation processing SRb (Figure 10) using the particle method (result of one cycle of calculation processing) performed on the dynamic three-dimensional head and neck particle model 50a in which the muscle activity rate value has been changed, and the muscle activity rate that can accurately reproduce the behavior of the hyoid bone 52 during swallowing is identified by analyzing the obtained evaluation function.
[0194] Here, there are two methods for the muscle activity analysis process for analyzing the muscle activity rate of the dynamic three-dimensional head and neck particle model 50a. The muscle activity analysis process according to the first method calculates an evaluation function based on the positions of muscle particles in the three-dimensional image during the swallowing simulation without considering the elapsed time and the speed of the muscle particles during the swallowing simulation using the dynamic three-dimensional head and neck particle model 50a, and obtains the optimal muscle activity rate for each muscle type from the analysis result of this evaluation function.
[0195] On the other hand, the muscle activity rate analysis process by the second method calculates an evaluation function by considering not only the positions of muscle particles in the three-dimensional image during the swallowing simulation using the dynamic three-dimensional head and neck particle model 50a, but also the elapsed time during the swallowing simulation and the speed of the muscle particles, and obtains the optimal muscle activity rate of the muscle particles for each muscle type from the analysis result of this evaluation function. The muscle activity rate analysis process by the first method and the muscle activity rate analysis process by the second method will be described in order below.
[0196] (7-1) <Muscle activity rate analysis processing by the first method> Fig. 13 is a flow chart showing the muscle activity rate analysis processing procedure by the above-mentioned first method. In this case, as shown in Fig. 13, the swallowing simulation device 1 moves from a start step to a subroutine SR1, performs a manual setting process, and then moves to the next subroutine SR2. The swallowing simulation device 1 performs a static optimization process in the subroutine SR2, and can finally identify the optimal muscle activity rate for each muscle type. Below, the manual setting process of the subroutine SR1 and the static optimization process of the subroutine SR2 will be described in order.
[0197] (7-1-1) <Manual setting process> In this case, in the swallowing simulation device 1, first, for each muscle type of the dynamic three-dimensional head and neck particle model 50a, the muscle activity rate α t (t indicates the time from the start of the swallowing simulation) is manually set by the developer based on his / her guess or experience.
[0198] In the swallowing simulation device 1, the developer or the like can manually set the optimal muscle activity rate α t When is set as an initial value in the dynamic three-dimensional head and neck particle model 50a, an analysis using a particle method is performed by the motion analysis unit 50, and a swallowing simulation that reproduces the motion of each head and neck organ of the dynamic three-dimensional head and neck particle model 50a from the start of swallowing to the end of swallowing is generated as the analysis result. As a result, the swallowing simulation device 1 can obtain a moving image that reproduces the rough basic motions of the skull 51a, 51b, hyoid bone 52, mandible 53, stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58 as a swallowing simulation from the start of swallowing to the end of swallowing.
[0199] In this embodiment, the process of repeating one cycle of calculation processing SRb by the particle method from the start of swallowing to the end of swallowing according to the above-mentioned "(5-2) <Structural analysis method of dynamic three-dimensional head and neck particle model incorporating only internal force>" and reproducing the movement of the dynamic three-dimensional head and neck particle model 50a from the start of swallowing to the end of swallowing is simply called swallowing simulation from the start of swallowing to the end of swallowing. Here, the moving image obtained as the analysis result of the swallowing simulation from the start of swallowing to the end of swallowing is composed of multiple frame images (also called three-dimensional images) arranged in chronological order from the start of swallowing to the end of swallowing.
[0200] Figure 14 shows the muscle activity rate α for each muscle type. t 13 is a schematic diagram showing dynamic three-dimensional head and neck particle models 50b1, 50b2, and 50b3 when any value of is manually set.
[0201] In FIG. 14, "50b1" is a dynamic three-dimensional head and neck particle model that appears in the 8th frame image of the generated moving image, which is obtained 0.27 S after the start of the swallowing simulation. "50b2" is a dynamic three-dimensional head and neck particle model that appears in the 40th frame image of the generated moving image, which is obtained 1.33 S after the start of the swallowing simulation. "50b3" is a dynamic three-dimensional head and neck particle model that appears in the 63rd frame image of the generated moving image, which is obtained 2.10 S after the start of the swallowing simulation. In addition, such frame images show the dynamic three-dimensional head and neck particle models 50b1, 50b2, and 50b3 that are stationary in the virtual three-dimensional space, and the dynamic three-dimensional head and neck particle models 50b1, 50b2, and 50b3 can be displayed from various angles by changing the viewpoint in the image.
[0202] During manual setting processing, the swallowing simulation device 1 first arbitrarily extracts multiple frame images from a moving image at a predetermined interval in accordance with an operation command input by the developer via the input unit 81. The swallowing simulation device 1 specifies the ideal position of the hyoid bone in each frame image based on a VF image or a 4DCT image obtained when the subject swallows, and draws a target image 52a at the ideal hyoid bone position specified in each frame image. The drawing of the target image 52a in each frame image is performed by the head and neck modeling unit 10.
[0203] For example, in the 8th frame image, the 40th frame image, and the 63rd frame image shown in FIG. 14, the position of the hyoid bone 52 is misaligned with the position of the target image 52a, suggesting that the behavior of the hyoid bone 52 in the swallowing simulation differs from the behavior of the hyoid bone when the subject actually swallows.
[0204] In the swallowing simulation device 1, the muscle activity rates α tBy changing the value of , the behavior of the stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58 changes during swallowing simulation, and the position of the hyoid bone 52 can be corrected in the virtual three-dimensional space within the frame image.
[0205] Therefore, the developer set muscle activity rates α t The value of is changed so that the position of the hyoid bone 52 approaches the position of the target image 52a in each frame image.
[0206] For example, in the eighth frame image obtained 0.27 S after the start of the swallowing simulation, when the position of the hyoid bone 52 is to be brought closer to the position of the target image 52a, the muscle activity rates α 0.27 Change the value of (muscle activity rate 0.27 S after the start of swallowing simulation).
[0207] In the swallowing simulation device 1, the muscle activity rates α of the stylohyoid muscles 56a and 56b, the geniohyoid muscle 57, and the sternohyoid muscle 58 are 0.27 A swallowing simulation (analysis by particle method) is performed 0.27 S after the start of the swallowing simulation using the dynamic three-dimensional head and neck particle model 50a in which the value of is changed. Specifically, one cycle of calculation processing SRb (FIG. 10) by the particle method is performed 0.27 S after the start of the swallowing simulation, and the positions of the stylohyoid muscles 56a, 56b, the geniohyoid muscle 57, and the sternohyoid muscle 58 at 0.27 S after the start of the swallowing simulation are calculated. As a result, the position of the hyoid bone 52 in the frame image of the eighth frame moves to positions corresponding to the positions of the stylohyoid muscles 56a, 56b, the geniohyoid muscle 57, and the sternohyoid muscle 58 at 0.27 S after the start of the swallowing simulation.
[0208] The developer judges whether or not the position of the hyoid bone 52 is misaligned with the position of the target image 52a based on the eighth frame image in which the target image 52a is drawn. In this manner, the developer adjusts the muscle activity rates α 0.27 Change in the value of and the muscle activity rate α 0.27 Using the dynamic three-dimensional head and neck particle model 50a with the value of changed, one cycle of calculation processing SRb (Figure 10) using the particle method is repeated 0.27 S from the start of the swallowing simulation until the position of the hyoid bone 52 approaches the position of the target image 52a.
[0209] Note that such processing is also performed on, for example, the 40th frame image 1.33 S after the start of the swallowing simulation and the 63rd frame image 2.10 S after the start of the swallowing simulation, which are extracted from the moving image.
[0210] Then, for each frame image extracted from the video, the muscle activity rate α t Once the value of is set, the muscle activation rate α t The muscle activity rate α between the frame images where the value of t For other frame images in which the value of has not been changed, the muscle activity rate α t Calculate the value of
[0211] For example, in the eighth frame image, the muscle activity rate α 0.27 The value of (muscle activity rate at 0.27 S from the start of the swallowing simulation) was set by manual setting processing, and the muscle activity rate α 1.33 When the value of (muscle activity rate at 1.33 S from the start of the swallowing simulation) is set by the manual setting process, the muscle activity rate α t is the muscle activity rate α0.27 and muscle activity rate α 1.33 A value calculated by linear interpolation from the values of and is set.
[0212] However, in such a manual setting process, the muscle activity rates α of the stylohyoid muscles 56a and 56b, the geniohyoid muscle 57, and the sternohyoid muscle 58 are t Since the value of is set based on the developer's intuition and inference, it is difficult to perfectly match the hyoid bone 52 and the target image 52a, so the developer performs manual setting processing until the hyoid bone 52 is brought to a certain degree close to the target image 52a.
[0213] Next, the above-mentioned manual setting process will be described with reference to the flowchart shown in Fig. 15. Fig. 15 is a flowchart showing the manual setting process procedure of the subroutine SR1 in Fig. 13. In this case, the swallowing simulation device 1 calculates the muscle activity rate α t A swallowing simulation is performed using dynamic three-dimensional head and neck particle model 50a with initial values set to, and a moving image is generated that is the analysis result of the swallowing simulation from the start of swallowing to the end of swallowing.In step S11, one frame image is extracted at a predetermined timing from the multiple frame images that make up the moving image, and the process proceeds to the next step S12.
[0214] In step S12, the swallowing simulation device 1 displays on the display unit 4 the VF images and 4DCT images of the head and neck organs obtained when the subject swallows, and presents the developer with the position of the hyoid bone when the subject swallows. In step S12, the swallowing simulation device 1 allows the developer to specify the ideal position of the hyoid bone in the frame images extracted in step S11 based on these VF images and 4DCT images, and allows the developer to draw a target image 52a indicating the ideal position of the hyoid bone in the frame images, and proceeds to the next step S13. Note that the drawing of the target image 52a in each frame image is performed by the head and neck modeling unit 10.
[0215] In step S13, the swallowing simulation device 1 calculates the muscle activity rate α t The developer modifies this and moves on to the next subroutine SR5.
[0216] At this time, the swallowing simulation device 1 calculates the muscle activity rate α set for each muscle type at the timing of each frame image extracted in step S11. t The value of the muscle activity rate α is displayed on the display unit 4, and the developer is informed of the muscle activity rate α t Check the value of muscle activity rate α t The value of is changed by the developer operating the input unit 81 to input a predetermined numerical value, and the value is changed by the muscle activity rate changing unit 80.
[0217] In subroutine SR5, the swallowing simulation device 1 performs one cycle of calculation processing SRb (FIG. 10) by the particle method at the time step (time) t of the frame image extracted in step S1 in accordance with the above-mentioned "(5-2)<Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", using the motion analysis unit 50, and then proceeds to the next step S14.
[0218] As a result, in the swallowing simulation device 1, one cycle of calculation processing SRb by the particle method is performed in the subroutine SR5, so that the position r t ´ and speed v t ' (step S9 in FIG. 10) can be obtained.
[0219] In step S14, when the particles (here, muscle particles, forced particles, and other particles) of the dynamic three-dimensional head and neck particle model 50a move by performing subroutine SR5, the developer or the like visually checks whether the amount of movement of all of these particles is small or not, and again determines whether to perform one cycle of calculation processing SRb using the particle method shown in subroutine SR5.
[0220] If the developer or the like judges that the amount of movement of all particles constituting the dynamic three-dimensional head and neck particle model 50a is large, it is determined that one cycle of calculation processing SRb by the particle method is necessary, and the process proceeds to the next step S15. In step S15, the swallowing simulation device 1 calculates the position r of the muscle particle i assumed at the time step (time) t in the next subroutine SR5. t and velocity v t (Step S1 in FIG. 10) is performed by dividing the position r of muscle particle i calculated in the previous subroutine SR5. t ´ and speed v t ' and the process returns to subroutine SR5.
[0221] In this way, the above-mentioned subroutine SR5, step S14, and step S15 are repeated until the developer, etc., judges that the movement amount of all particles constituting the dynamic three-dimensional head and neck particle model 50a has become small based on the subjective judgment of the developer, etc. When the developer, etc., judges that the movement amount of all particles constituting the dynamic three-dimensional head and neck particle model 50a has become small based on the subjective judgment of the developer, etc., one cycle of calculation processing SRb by the particle method is not necessary, and the process proceeds to the next step S16.
[0222] In step S16, the developer receives a muscle activity rate α t A frame image in which the position of the hyoid bone 52 has been moved by changing the value of is presented. In step S16, the developer visually checks whether the position of the hyoid bone 52 and the position of the target image 52a are similar to each other in the frame image in which the target image 52a is drawn, and determines whether the muscle activity rate α t It is determined whether the optimization setting is set.
[0223] If a negative result is obtained in step S16, this indicates that there is a large deviation between the position of the hyoid bone 52 and the position of the target image 52a in the frame image in which the target image 52a is drawn, and in this case, the above-mentioned processing of step S13, subroutine SR5 and steps S14 to S16 is repeated until the position of the hyoid bone 52 approaches the position of the target image 52a in the frame image.
[0224] On the other hand, if a positive result is obtained in step S16, this indicates that the position of the hyoid bone 52 in the frame image has come closer to the position of the target image 52a to a certain extent, and at this time, the manual setting process procedure described above is terminated. Such a manual setting process procedure is performed for each frame image extracted from the video. Then, when the manual setting process has been performed for all frame images extracted at a predetermined timing from the video, the muscle activity rates α t The muscle activity rate α determined by the manual setting process is t Linear interpolation is performed using
[0225] For example, FIG. 16 shows the muscle activity rate α m t (In this notation, m indicates muscle type and t indicates time) is a table showing an example. Here, two muscle types, muscle 1 and muscle 2, are shown as examples, and frame images extracted from video images that are analysis results of the swallowing simulation are used as "key frames." In FIG. 16, key frame "1." is a frame image 0.1 S after the start of the swallowing simulation, key frame "2." is a frame image 0.4 S after the start of the swallowing simulation, and key frame "3." is a frame image 0.7 S after the start of the swallowing simulation. For example, the muscle activity rate α between 0.1 S and 0.4 S is t For example, α1 of 0.1S 0.1 And the 0.4S α1 0.4 It is calculated by linear interpolation with
[0226] (7-1-2) <Static optimization processing> Next, the static optimization process of subroutine SR2 in the muscle activity rate analysis process procedure shown in Fig. 13 will be described below. In the static optimization process, first, the muscle activity rate α set in the dynamic three-dimensional head and neck particle model 50a of a predetermined frame image by the manual setting process is used. Here, for convenience of explanation, the muscle activity rate manually optimized for the dynamic three-dimensional head and neck particle model 50a of a certain frame image by the above-mentioned manual setting process is expressed by the following formula (24).
number
[0227] Muscle activity rate α1 k,0 ,…,α n k,0 The subscripts 1 to n are identifiers for identifying the muscle type m of the head and neck organs (m = 1, 2, ..., n). In the above formula (24), the muscle activity rate α1 k,0 ,…,α n k,0 In the case of the dynamic three-dimensional head and neck particle model 50a shown in FIG. 11, four muscle types, namely, the stylohyoid muscles 56a and 56b, the geniohyoid muscle 57, and the sternohyoid muscle 58, are set, so the muscle activity rate of the above formula (24) is k,0 ,…,α4 k,0 It becomes.
[0228] Muscle activity rate α k,0 The superscript k in indicates the number of optimization calculations (to be described later) of the evaluation function calculated by the muscle activity rate analysis unit 91. k,0 The superscript 0 in the above table is an identifier that identifies the changed state of the muscle activity rate value set for each muscle type, and the notation 0 indicates that the muscle activity rate value is the initial value, that is, the value has not been changed.
[0229] The static optimization process is carried out by using the muscle activity rate α k,0 Whether the behavior of the head and neck organs during swallowing of the subject is faithfully reproduced is evaluated using the evaluation function shown in the following equation (25).
number
[0230] e k,0 The superscript k in the formula (25) indicates the number of optimization calculations (to be described later) of the evaluation function calculated by the muscle activity rate analysis unit 91, and k=1, 2, ..., k max The superscript 0 indicates that this is an evaluation function for the initial value where no changes have been made to the muscle activity rate values set for the muscle particles for each muscle type. Os indicates a set of multiple head and neck organs with muscle activity rates set in the dynamic three-dimensional head and neck particle model 50a. Oj indicates the head and neck organ for which the evaluation function is calculated among the multiple head and neck organs for which the muscle activity rate is set. i∈Oj 1 is the number of particles in the head and neck organs Oj. w P,Oj is a position r in the head and neck organ (Oj) for which the evaluation function is calculated. i k,0 This constant w P,Oj is determined empirically. For example, 10 5 ~10 7 is desirable.
[0231] r i k,0 is the analysis result by the particle method in the motion analysis unit 50 obtained when calculating the evaluation function for the kth optimization calculation, and indicates the position of muscle particle i in the three-dimensional image at a predetermined time during swallowing. The subscript i is an identifier that identifies muscle particle i. The superscript 0 indicates that muscle particle i is in the initial value state where no changes have been made to the muscle activity rate value set for the muscle particle for each muscle type.
[0232] s i indicates a target position of the muscle particle i in the three-dimensional image at a predetermined time during swallowing. Such a target position of the muscle particle i is specified in advance based on anatomical knowledge, VF images, 4DCT images, etc. when the subject swallows, and is stored in the storage unit 83.
[0233] The target position s of muscle particle i from VF images, 4DCT images, etc. when the subject swallows i The identification of the target head and neck organ is, for example, performed by tracing the target head and neck organ and analyzing the movement of the head and neck organ. Since the muscle particle i is a part of the targeted head and neck organ, the target position s i can be determined from the motion of the head and neck organs. Note that the motion of the head and neck organs can be determined solely from medical knowledge, and the target position s of muscle particle i can be calculated from the motion of the head and neck organs. i may be determined.
[0234] w α is a constant that weights the muscle activity rate. α is determined empirically. For example, 0~w P,Oj ×10 -8 is desirable.
[0235] α m k,0 indicates the muscle activity rate of muscle particle i in muscle type m used when calculating the evaluation function for the kth optimization calculation. The subscript m is an identifier defined for each muscle type to identify the muscle type of the head and neck organs, and is 1 to n in the above formula (24).
[0236] The number of optimization calculations k is determined by the search direction ya k This is the number of times that the final evaluation function is determined by performing the determination of the evaluation function and the line search, and is preset in the optimization number setting unit 97 of the swallowing simulation device 1. In this embodiment, the first optimization calculation number k of the static optimization process that is first performed after the manual setting process is completed is set to 1, and the maximum number of optimization calculation numbers k that is preset is set to k. max where k=1,…,k max Let us assume that.
[0237] The position r of this muscle particle i i k,0 can be obtained by one cycle of calculation processing SRb (FIG. 10) by the particle method at time t. In FIG. 10, the position rt In this embodiment, the symbol indicating the position of muscle particle i is expressed as r i k,0 , r t However, the position r of these muscle particles i i k,0 , the position r of muscle particle i t The numerical values indicated by ´ etc. are the same, for example, r i k,0 =r t ´.
[0238] The evaluation function calculation unit 92 of the swallowing simulation device 1 calculates a constant w P,Oj and the constant w α and the target position s of muscle particle i i and the muscle activity rate α obtained by the manual setting process. k,0 (k=1, α 1,0 ) from the storage unit 83, and the muscle activity rate α k,0 The position r of muscle particle i in the 3D image at i k,0 Based on the above (25), the evaluation function e k,0 Calculate.
[0239] In addition, the evaluation function e used in the static optimization process is k,0 The general formula is given by the following formula (26).
number
[0240] The superscript Ph is an identifier for identifying the change state of the muscle activity rate when calculating the evaluation function while changing the value of the muscle activity rate set for muscle particle i for each specified muscle type. For example, when the muscle activity rate is in the initial value state without being changed, "Ph" is "0", and e k,0 , r i k,0 , α m k,0When the value of the muscle activity rate of a specific muscle type m is changed, "Ph" becomes "▽;m" (m indicates the muscle type, 1 to n), and e k,▽;m , r i k,▽;m , α m k,▽;m In addition, the search direction ya k When determining and searching the line, "Ph" becomes "p" and e k,p , r i k,p , α m k,p Furthermore, the search direction ya k When the muscle activation rate is finally optimized after the kth optimization calculation, "Ph" becomes "pend" and e k,pend , r i k,pend , α m k,pend When "k" is the maximum number of times "kmax", e kmax,pend , r i kmax,pend , α m kmax,pend It becomes.
[0241] Next, the static optimization process using the above-mentioned formulas (25) and (26) will be described with reference to the flowcharts shown in Figures 17A and 17B. Figures 17A and 17B are flowcharts showing the static optimization process procedure of subroutine SR2 in Figure 13. In this case, when the manual setting process ends, the swallowing simulation device 1 selects one frame image from the multiple frame images whose muscle activity rates have been optimized in the manual setting process. The static optimization process of this embodiment is performed on all frame images extracted from the moving image in the manual setting process, but the following description will focus on one frame image first.
[0242] In step S21, the swallowing simulation device 1 (i) calculates the muscle activity rate α set by the manual setting process in the selected frame image. k,0 is read from the storage unit 83, and the muscle activity rate α k,0Then, the process moves to the next subroutine SR6.
[0243] In the first subroutine SR6, the position r of the muscle particle i assumed in the dynamic three-dimensional head and neck particle model 50c at the time step (time) t of the frame image is calculated. t and velocity v t (Step S1 in FIG. 10) and the muscle activity rate α set in step S21 k,0 Using these, in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", one cycle of calculation processing SRb using the particle method shown in Figure 10 is performed by the motion analysis unit 50, and then the process proceeds to the next step S22.
[0244] As a result, in the swallowing simulation device 1, one cycle of calculation processing SRb by the particle method is performed in the subroutine SR6, so that the position r t ´ and speed v t ' (step S9 in FIG. 10). Here, the position r of muscle particle i obtained in subroutine SR6 can be calculated. t ´ is r i k,0 and the velocity v of muscle particle i calculated in subroutine SR6. t ´ is v i k,0 It is written as follows.
[0245] In step S22, the swallowing simulation device 1 (i) calculates the muscle activity rate α k,0 and (ii) the muscle activation rate α k,0 The position r obtained from the dynamic three-dimensional head and neck particle model 50a by one cycle of calculation processing SRb using the particle method i k,0 Using the above equation (25), the evaluation function e k,0 Then, the process proceeds to step S23.
[0246] In step S23, the swallowing simulation device 1 calculates the velocity v of the muscle particle i calculated in the subroutine SR6. i k,0and the evaluation function e calculated in step S22 k,0 Based on this, it is determined whether or not a steady state has been reached.
[0247] Specifically, the swallowing simulation device 1, by the steady state determination unit 93, (i) determines the velocities v of all muscle particles i obtained in the subroutine SR6. i k,0 is equal to or less than a threshold value (for example, equal to or less than 5.0 mm / s, preferably equal to or less than 0.1 mm / s), and (ii) the evaluation function e k,0 When it is determined that the amount of change (rate of change) of is equal to or less than a threshold (for example, 1% or less, preferably 0.01% or less), it is determined that the steady state is being reached.
[0248] Here, the evaluation function e k,0 The determination of whether the change in the latest evaluation function e obtained in step S22 is equal to or smaller than the threshold value is made. k,0 and the past evaluation function e obtained in the previous step S22. k,0 For example, the latest evaluation function e calculated in step S22 is k,0 is denoted as "e_current", and the past evaluation function e k,0 If we write it as "e_previous", the evaluation function e k,0 The amount of change can be calculated using the following formula. (e_current - e_previous) / e_previous
[0249] In this case, the swallowing simulation device 1 calculates the evaluation function e k,0 It is determined whether the change in the evaluation function e in step S22 is equal to or less than the threshold value. k,0 If this is the first calculation, in step S23, the evaluation function e k,0 Since the amount of change cannot be calculated, the process moves to step S24.
[0250] In this embodiment, (i) the velocity v of muscle particle i i k,0 and (ii) the evaluation function ek,0 However, the present invention is not limited to this. For example, (ii) evaluation function e k,0 Alternatively, it may be possible to determine whether or not the steady state is being reached by using only the above-mentioned. The same applies to other "determinations of the steady state" described later.
[0251] If a negative result is obtained in step S23, this means that the steady state has not been reached or that the evaluation function e k,0 This is the first calculation of the evaluation function e k,0 In this case, the swallowing simulation device 1 proceeds to the next step S24. In step S24, the swallowing simulation device 1 calculates the change amount of the position r of the muscle particle i assumed at the time step (time) t in the next subroutine SR6. t and velocity v t (Step S1 in FIG. 10) is calculated by the subroutine SR6. i k,0 and velocity v i k,0 and the process returns to subroutine SR6.
[0252] In subroutine SR6, the position r of muscle particle i updated in step S24 is i k,0 and velocity v i k,0 and the muscle activity rate α set in step S21. k,0 Using these, the motion analysis unit 50 again performs one cycle of calculation processing SRb using the particle method shown in Figure 10 in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", and proceeds to the next step S22.
[0253] In the second subroutine SR6, the position r i k,0 is the position r of muscle particle i assumed in step S1 of Fig. 10. t and velocity v tThe position r calculated in the first subroutine SR6 i k,0 and velocity v i k,0 Since the calculation process SRb is performed by replacing the position r i k,0 will be different from the value.
[0254] In step S22, the swallowing simulation device 1: (i) the muscle activity rate α k,0 (ii) the position r newly calculated by one cycle of calculation processing SRb using the particle method in the second subroutine SR6 i k,0 ((Point r in step S9 in FIG. 10) t Using these, we again use the evaluation function e k,0 Then, the process proceeds to step S23.
[0255] In this manner, in the swallowing simulation device 1, the position r of the muscle particle i used in one cycle of the calculation process SRb by the particle method in the subroutine SR6 is kept constant until the steady state is determined in step S23. t speed v t (Step S1 in FIG. 10) is performed by the subroutine SR6. i k,0 and velocity v i k,0 While updating the value each time, the above-mentioned subroutine SR6 and steps S22 to S24 are repeatedly performed.
[0256] If a positive result is obtained in step S23, this indicates that a steady state has been reached, and at this time, the swallowing simulation device 1 proceeds to the next step S30.
[0257] In step S30, the swallowing simulation device 1 uses the muscle activity rate α k,0 Muscle activation rate α with different values k,▽;m Then, the process moves to the next subroutine SR7. Here, the initial value of the muscle activity rate αk,0 Muscle activation rate α with different values k,▽;m is the muscle activity rate α as shown in the following equation (27). k,▽;m The superscript ▽;m is an identifier that indicates the muscle type m whose value has been changed.
number
[0258] Muscle activity rate α k,▽;m For example, as shown in the above equation (27), the muscle activity rate α of a given muscle type m is m k,0 The value is changed by adding h to the value. h may be a preset value or may be a value set by the developer each time, but it is preferable that h is a small value such as 0.0001 to 0.01.
[0259] Muscle activity rate α to add h m k,0 For example, muscle activity rate α1 k,0 ,…,α n k,0 Among them, the muscle activity rate α of any one muscle type m m k,0 Add h to m. First, let m = 1 and the muscle activity rate α1 k,0 Muscle activity rate α with h added k,▽;1 (α1 k,0 +h,α2 k,0 ,…,α n k,0 ) is used. k,▽;m The value of is changed by the muscle activity rate changing unit 80.
[0260] In the first subroutine SR7, the position r of the muscle particle i when it is determined in step S23 that the steady state is i k,0 and velocity v i k,0 and the muscle activity rate α changed in step S30 k,▽;m (When m=1, muscle activity rate α k,▽;1) and again according to the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", one cycle of calculation processing SRb using the particle method shown in Figure 10 is performed by the motion analysis unit 50, and then the process proceeds to the next step S31.
[0261] As a result, in the swallowing simulation device 1, one cycle of calculation processing SRb using the particle method is performed in the subroutine SR7, thereby obtaining the muscle activity rate α k,▽;m The position r of muscle particle i when t ´ and speed v t ' (step S9 in FIG. 10). Here, the position r of muscle particle i obtained in subroutine SR7 can be calculated. t ´ is r i k,▽;m (When m=1, position r k,▽;1 ) and the velocity v of muscle particle i calculated in subroutine SR7. t ´ is v i k,▽;m (When m=1, the velocity v k,▽;1 ) is written as
[0262] In step S31, the swallowing simulation device 1 (i) changes the muscle activity rate α k,▽;m (When m=1, muscle activity rate α k,▽;1 ) and (ii) muscle activation rate α k,▽;m The position r obtained from the dynamic three-dimensional head and neck particle model 50a by one cycle of calculation processing SRb using the particle method i k,▽;m (When m=1, position r i k,▽;1 ) and, using the above equation (26), the evaluation function e k,▽;m (When m=1, the evaluation function e k,▽;1 ) and proceed to the next step S32.
[0263] In step S32, the swallowing simulation device 1 calculates the velocity v of the muscle particle i calculated in the subroutine SR7. i k,▽;m (When m=1, the velocity v k,▽;1) and the evaluation function e obtained in step S31 k,▽;m (When m=1, the evaluation function e k,▽;1 The determination in step S32 as to whether or not the steady state has been reached is the same as that in step S23 described above, and the steady state determination unit 93 determines whether or not the steady state has been reached based on (i) the velocities v of all muscle particles i obtained in the subroutine SR7. i k,▽;m is equal to or less than a threshold value (for example, equal to or less than 5.0 mm / s, preferably equal to or less than 0.1 mm / s), and (ii) the evaluation function e k,▽;m is determined to be equal to or less than a threshold (for example, 1% or less, preferably 0.01% or less), it is determined to be in a steady state.
[0264] In addition, since the determination of the steady state in step S32 is performed by the same determination method as in step S23 as described above, the evaluation function e k,▽;m If this is the first calculation, in step S32, the evaluation function e k,▽;m Since the amount of change cannot be calculated, the process moves to step S33.
[0265] If a negative result is obtained in step S32, this means that the steady state has not been reached or that the evaluation function e k,▽;m This is the first calculation of the evaluation function e k,▽;m In this case, the swallowing simulation device 1 proceeds to the next step S33. In step S33, the swallowing simulation device 1 calculates the amount of change in the position r of the muscle particle i used as the initial value in the first subroutine SR7. i k,0 and velocity v i k,0 (In step S1 of FIG. 10, t and velocity v t ) is calculated by the position r i k,▽;1 and velocity v i k,▽;1 Then, the process returns to subroutine SR7.
[0266] In subroutine SR7, the new position r of muscle particle i updated in step S33 is i k,▽;1 and velocity v i k,▽;1 and the muscle activity rate α set in step S30 k,▽;1 Using these, the motion analysis unit 50 again performs one cycle of calculation processing SRb using the particle method shown in Figure 10 in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", and then proceeds to the next step S31.
[0267] In step S31, the swallowing simulation device 1 (i) calculates the muscle activity rate α k,▽;1 (ii) the position r newly calculated by one cycle of calculation processing SRb using the particle method in the second subroutine SR7 i k,▽;1 ((Point r in step S9 in FIG. 10) t Using the above equation (26), we again obtain the evaluation function e k,▽;1 Then, the process proceeds to step S32.
[0268] In this manner, in the swallowing simulation device 1, the position r of the muscle particle i used in one cycle of the calculation process SRb by the particle method in the subroutine SR7 is determined to be in a steady state in step S32. t speed v t (Step S1 in FIG. 10) is performed by the subroutine SR7. i k,▽;1 and velocity v i k,▽;1 While updating the value each time, the above-mentioned subroutine SR7 and steps S31 to S33 are repeatedly performed.
[0269] If a positive result is obtained in step S32, this indicates that a steady state has been reached, and at this time, the swallowing simulation device 1 proceeds to the next step S34.
[0270] In step S34, the swallowing simulation device 1 calculates the muscle activity rate α k,▽;m It is determined whether the change in muscle activity rate α in step S30 has been performed for all muscle types m (m=1, 2, ..., n), i.e., whether m=n. If a negative result is obtained in step S34, this means that the change in muscle activity rate α in step S30 k,▽;m This indicates that the change in muscle activity rate α has not been made for all muscle types m, and in this case, the process proceeds to the next step S35. k,▽;1 If set, other muscle activity rates α k,▽;2 ,…,α k,▽;n Since it has not been set in step S30, a negative result is obtained in step S34.
[0271] In step S35, the swallowing simulation device 1 calculates the muscle activity rate α k,▽;m m is changed to m+1, and muscle activity rate α k,▽;m+1 Then, m+1 is regarded as m, and the muscle activity rate α k,▽;m Then, the process returns to step S30. In step S30, the swallowing simulation device 1 determines whether the muscle activity rate α k,▽;1 (α1 k,0 +h,α2 k,0 ,…,α n k,0 ) is set, the muscle activity rate α k,▽;2 (α1 k,0 ,α2 k,0 +h,…,α n k,0 ) and proceed to the next subroutine SR7.
[0272] As a result, for example, the muscle activity rate α set in step S30 k,▽;2 Regarding the muscle activity rate α k,▽;1 Similarly, the above-mentioned subroutine SR7 and steps S31 to S33 are repeated until a steady state is determined in step S32.
[0273] In step S30, the muscle activity rate α k,▽;2Even if the above is set, in the first subroutine SR7, the position r of the muscle particle i when it is determined to be in a steady state in step S23 is set as i k,0 and velocity v i k,0 and the muscle activity rate α set in step S30 k,▽;2 Using these, one cycle of calculation processing SRb using the particle method shown in Figure 10 is performed by the motion analysis unit 50 in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>".
[0274] In addition, in step S30, the muscle activity rate α k,▽;3 Even if the above is set, in the first subroutine SR7, the position r of the muscle particle i when it is determined to be in a steady state in step S23 is set as i k,0 and velocity v i k,0 and the muscle activity rate α set in step S30 k,▽;3 Using these, one cycle of calculation processing SRb using the particle method shown in Figure 10 is performed by the motion analysis unit 50 in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>".
[0275] If a positive result is obtained in step S34, this means that the muscle activity rate α k,▽;m The change in the evaluation function e has been made for all muscle types m, that is, the evaluation function e determined to be in a steady state for all muscle types m in step S32 k,▽;1 ,e k,▽;2 ,…,e k,▽;n is obtained, and at this time, the swallowing simulation device 1 proceeds to step S40 shown in FIG. 17B.
[0276] The swallowing simulation device 1 calculates the evaluation function e that is determined to be in a steady state for all muscle types m. k,▽;m (e k,▽;1 ,e k,▽;2 ,…,e k,▽;n ) is obtained, this muscle activity rate α k,▽;mThe search direction ya is determined and a line search is performed using
[0277] Specifically, in step S40, the swallowing simulation device 1 performs the evaluation function e k,0 , e k,▽;m Using the above, the gradient ▽e is calculated based on the following equations (28) and (29). k Then, the process proceeds to step S41.
number
[0278] (∂e / ∂α m ) k is expressed as the following equation (29). m is α1,…,α n where m is an identifier for identifying the muscle type. k,0 indicates the evaluation function determined to be in a steady state in step S23, and e k,▽;m (e k,▽;1 ,e k,▽;2 ,…,e k,▽;n ) indicates the evaluation function that is determined to be in a steady state in step S32.
number
[0279] In addition, such a gradient ▽e k The calculation of is performed by the gradient calculation unit 94 of the swallowing simulation device 1.
[0280] In step S41, the swallowing simulation device 1 determines the search direction ya m k (m is muscle type, ya k (=ya1 k ,…,ya n k Then, the process proceeds to step S42. Here, the search direction ya kThe determination of can be performed, for example, by the conjugate gradient method, but the search direction ya can be determined by various other methods. k The search direction ya may be determined by the conjugate gradient method. k When determining the search direction ya, the calculation method is different when the optimization calculation number k is the first time (k=1) and when the optimization calculation number k is the second time or more (k=2 or more). When the optimization calculation number k=1, the search direction ya k can be determined by the following equation (30):
number
[0281] On the other hand, when the optimization calculation count k is 2 or more, the search direction ya k can be determined by the following equation (31). That is, the search direction ya k If the determination of and line search (described later) have not been performed even once, the optimization calculation count k=1, and the above formula (30) is used. On the other hand, if the search direction ya k When the determination and the line search (described later) have been performed once, the number of optimization calculations is k=2 or more, and the following formula (31) is used.
number
[0282] In addition, such a search direction ya k The determination of is performed by the search direction determination unit 95 of the swallowing simulation device 1. Note that, "β k " is different from "β", which indicates the degree of relaxation described later.
[0283] As described above, in step S41, the search direction ya k When is determined, the swallowing simulation device 1 then performs a line search. The line search first calculates the muscle activity rate α k,p (α m k,p (=α1 k,p ,…,α nk,p The superscript p indicates the search direction ya k This is an identifier that indicates that the muscle activity rate α is being determined and a line search is being performed. k,0 is α m k,0 (=α1 k,0 ,…,α n k,0 ) can also be written as That is, the following equation (32) is m k,p = α m k,0 -μ p ·ya m k It can also be written as:
number
[0284] μ p is a step size, which is an arbitrary value that can be changed and is a numerical value that is increased by the developer when performing a line search. In step S42, the swallowing simulation device 1 adds a predetermined value of the step size μ p Then, the process proceeds to step S43. Here, as an example, p =h(p+0.015p 3 ) (p in parentheses indicates a predetermined value). Note that such a step size μ p The setting is performed by the line search processing unit 96.
[0285] In step S43, the swallowing simulation device 1 calculates the search direction ya k and the step size μ set in step S42 p and muscle activity rate α k,0 Using the above equation (32), the muscle activity rate α k,p Then, the process moves to the next subroutine SR8. k,p The calculation of is performed by the line search processor 96.
[0286] In the subroutine SR8, the swallowing simulation device 1 calculates the position r of the muscle particle i when it is determined in step S23 that the state is steady. i k,0 and velocity v i k,0 and the muscle activity rate α calculated in step S43. k,p Using these, the motion analysis unit 50 again performs one cycle of calculation processing SRb using the particle method shown in Figure 10 in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", and proceeds to the next step S44.
[0287] In step S44, the swallowing simulation device 1 (i) calculates the muscle activity rate α k,p and (ii) the position r calculated by one cycle of calculation processing SRb using the particle method in subroutine SR8. i k,p Using the above equation (26), the evaluation function e k,p Then, the process proceeds to step S45.
[0288] In step S45, the swallowing simulation device 1 calculates the velocity v of the muscle particle i calculated in the subroutine SR8. i k,p and the evaluation function e calculated in step S44 k,p The determination of whether or not the steady state has been reached in step S45 is the same as that in step S23 described above, and the steady state determination unit 93 determines whether or not the steady state has been reached based on (i) the velocity v of the muscle particle i calculated in the subroutine SR8, i k,p is equal to or less than a threshold value (for example, equal to or less than 5.0 mm / s, preferably equal to or less than 0.1 mm / s), and (ii) the evaluation function e k,p is determined to be equal to or less than a threshold value (for example, 1% or less, preferably 0.01% or less), it is determined to be in a steady state.
[0289] If a negative result is obtained in step S45, this indicates that the steady state has not been reached, and in this case, the swallowing simulation device 1 proceeds to the next step S46. In step S46, the swallowing simulation device 1 calculates the position r of the muscle particle i assumed at the time step (time) t in the next subroutine SR8. i k,0 and velocity v i k,0 (In step S1 of FIG. 10, t and velocity v t ) is calculated by the subroutine SR8. i k,p and velocity v i k,p Then, the process returns to subroutine SR8.
[0290] In subroutine SR8, the new position r of muscle particle i updated in step S46 is i k,p and velocity v i k,p and the muscle activity rate α calculated in step S43. k,p Using the above, according to the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", one cycle of calculation processing SRb by the particle method shown in FIG. 10 is performed by the motion analysis unit 50, and a new position r of the muscle particle i is obtained. i k,p and velocity v i k,p (In step S9 of FIG. 10, t ´ and speed v t ') is calculated.
[0291] In this manner, in the swallowing simulation device 1, the processes of the above-mentioned subroutine SR8 and steps S44 to S46 are repeated until a steady state is determined in step S45.
[0292] If a positive result is obtained in step S45, this indicates that a steady state has been reached, and at this time, the swallowing simulation device 1 proceeds to the next step S47.
[0293] In step S47, the swallowing simulation device 1 determines that the evaluation function e k,p It is determined whether or not the condition of the following equation (33) is satisfied. Note that this determination is made by the steady state determination unit 93.
[0294]
number
[0295] In the above formula (33), e k,(p+1) is the evaluation function e k,p After calculating, the step size μ p Increase the value of step size μ p+1 Specifically, the evaluation function e k,p After calculating, the process proceeds to step S48, where the step size μ p p is set to p+1 and the step size μ p+1 Then, p+1 is regarded as p, and the step size μ p In step S42, the swallowing simulation device 1 calculates the step size μ p Step size μ p+1 Step size μ p In step S43, the muscle activity rate α is calculated from the above equation (32). k,(p+1) Request.
[0296] Thereafter, in the swallowing simulation device 1, the processes of the above-mentioned subroutine SR8 and steps S44 to S46 are repeated until a steady state is determined in step S45, and the evaluation function e k,(p+1) Using the previously calculated evaluation function e k,p It is judged whether or not satisfies the condition of the above formula (33).
[0297] If a positive result is obtained in step S47, this means that the evaluation function e calculated in step S44 k,p indicates that the condition of the above formula (33) is satisfied, and at this time, the swallowing simulation device 1 proceeds to the next step S49. Note that in step S47, the evaluation function e k,p is the evaluation function finally obtained after the optimization calculation count k. Hereinafter, the evaluation function e k,pend and the evaluation function e k,pend The muscle activity rate at this time is called muscle activity rate α k,pend It is written as follows.
[0298] In step S49, the swallowing simulation device 1 determines whether the optimization calculation count k, which indicates the number of times the above-mentioned subroutines SR6 to S47 have been performed on one frame image, is equal to or greater than the maximum number k preset in the optimization count setting unit 97. max Determine whether it has been reached.
[0299] If a negative result is obtained, this means that the number of optimization calculations k, which is the number of processes performed in the above-mentioned subroutines SR6 to S47, has reached the maximum number k max In this case, the swallowing simulation device 1 advances to step S50. In step S50, the swallowing simulation device 1 increases the muscle activity rate α k,pend Set k to k+1 (for example, when the optimization calculation number is k=1 (first time), the next optimization calculation number is k=2 (second time)), set pend to 0, and set the muscle activation rate α k+1,0 Toshi (α k+1,0 ←α k,pend ), and muscle activity rate α k+1,0 Consider k+1 as k and calculate muscle activity rate α k,0 Then (k←k+1), and the process returns to subroutine SR6.
[0300] Specifically, in the subroutine SR6, in step S47, an evaluation function e that satisfies the condition of the above formula (33) is k,pend The muscle activity rate α used to calculatek,pend muscle activity rate α k,0 and this muscle activity rate α k,0 and the initial position r of muscle particle i assumed as the initial value at time step t of the frame image. t and velocity v t Using this (step S1 in FIG. 10), in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", one cycle of calculation processing SRb using the particle method shown in FIG. 10 is performed by the motion analysis unit 50, and then the next step S22 is performed.
[0301] In the above embodiment, in the subroutine SR6 performed after step S50, the initial position r of the muscle particle i assumed as the initial value at the time step (time) t of the frame image is t and velocity v t is used to perform one cycle of calculation processing SRb using the particle method, but the present invention is not limited to this.
[0302] For example, in the subroutine SR6 executed after step S50, when the number of optimization calculations k, which will be described later, is k=2 or more, α k-1,pend (That is, in the previous optimization calculation count k, the evaluation function e that satisfies the condition of the above formula (33) in step S47 is k,pend The muscle activity rate α used to calculate k,pend ) may use the position and velocity when the steady state is reached. This eliminates the need for the calculation process in subroutine SR6.
[0303] After that, the above-mentioned steps S22 to S47 are performed to obtain the muscle activity rate α k,pend (For example, muscle activity rate α 2,pend The swallowing simulation device 1 repeats the process of the subroutine SR6 to step S47 until a positive result is obtained in step S49, that is, until the maximum number of times k max This is repeated until the evaluation function e kmax,pend Then, the evaluation function e kmax,pend Muscle activation rate α kmax,pendLet be the final muscle activation rate of the static optimization process.
[0304] The swallowing simulation device 1 finally calculates the evaluation function e kmax,pend and muscle activity rate α kmax,pend When the optimal muscle activity rate α is calculated, the static optimization process for one frame image is terminated. In this embodiment, the static optimization process is performed by manually setting the optimal muscle activity rate α k,0 is performed for all set frame images.
[0305] In addition, the static optimization process is performed for each frame image that is extracted from the moving image that is the analysis result of the swallowing simulation and that has been subjected to the manual setting process. However, for frame images that have not been subjected to the static optimization process in the moving image that is the analysis result of the swallowing simulation from the start of swallowing to the end of swallowing, the two muscle activity rates α kmax,pend The value of the muscle activity rate α is calculated and set by linear interpolation from the above.
[0306] As a result, the swallowing simulation device 1 can calculate the optimal muscle activity rate α kmax,pend It is possible to set the above, and to reproduce the movement of the head and neck organs during swallowing more accurately than ever before.
[0307] FIG. 18 shows the muscle activity rate α finally obtained by the static optimization procedure described above. m t Figure 18 shows an example of muscle activity rate α kmax,pend represents the muscle activity rate α kmax,pend α m t The muscle activity rate α m t m is an identifier for identifying muscle types, and in this case, muscle 1 and muscle 2 are identified. Also, the muscle activation rate α m t The time steps (times) t for which the above is calculated are 0.1S, 0.2S, 0.3S, and 0.4S as examples.
[0308] (7-2) <Muscle activity analysis processing by the second method> Next, the muscle activity analysis process by the second method will be described below. The muscle activity analysis process by the second method considers not only the positions of the muscle particles in the three-dimensional image during the swallowing simulation using the dynamic three-dimensional head and neck particle model 50a, but also the elapsed time and the speed of the muscle particles during the swallowing simulation, and obtains the optimal muscle activity rate for each muscle type.
[0309] Fig. 19 is a flow chart showing the muscle activity rate analysis processing procedure by the second method. As shown in Fig. 19, in the muscle activity rate analysis processing procedure by the second method, similarly to the muscle activity rate analysis processing procedure by the first method shown in Fig. 13, the process proceeds from a start step to a subroutine SR1 where a manual setting process is performed, and then the process proceeds to the next subroutine SR2 where a static optimization process is performed.
[0310] In the muscle activity rate analysis procedure according to the second method, a static optimization process is performed in subroutine SR2, and then a dynamic optimization process is performed in subroutine SR3. Finally, the muscle activity rate α calculated in the dynamic optimization process is kmax,pend This differs from the muscle activity rate analysis processing procedure according to the first method described above in that the optimal muscle activity rate is set to be the muscle activity rate.
[0311] Here, the manual setting process of subroutine SR1 and the static optimization process of subroutine SR2 are the same as those described above in “(7-1) <Muscle activity rate analysis process using the first method>”, so the following explanation focuses on the dynamic optimization process of subroutine SR3.
[0312] Fig. 20 is a flowchart showing the procedure of the dynamic optimization process shown in Fig. 19. As shown in Fig. 20, in the swallowing simulation device 1, the process proceeds from the start step to step S60, and the optimization target time ts~(ts+Δt opt ) and proceed to the next step S61. Here, in step S60, the optimization start time is set to ts, and the optimization end time is set to ts+Δtopt The case where the optimization start time ts is set as follows is described below. The optimization start time ts is set from the start of the swallowing simulation, and for example, the time t from the start of the swallowing simulation (for example, time 1.0S) is set as the optimization start time ts (time 1.0s).
[0313] In step S61, the swallowing simulation device 1 selects the optimization end time (ts+Δt opt ) Muscle activation rate α ts+△topt (Note that here, the muscle activity rate α kmax,pend At time (ts+△t opt ) and α ts+△topt (i.e., α kmax,pend = α ts+△topt Then, the process proceeds to step S62. Here, the muscle activity rate α ts+△topt is the end time of optimization (ts+△t opt ), but when expressed as the time t from the start of the swallowing simulation, the muscle activity rate α t+△topt It can also be expressed as:
[0314] In step S62, the swallowing simulation device 1 determines the optimization end time (ts+Δt opt ) Muscle activation rate α ts+△topt The optimization end time of the dynamic optimization process (ts+△t opt ) Initial muscle activity rate α k,0 (i.e., α ts+△topt = α k,0 ), and then the process moves to the analysis process of the next subroutine SR10.
[0315] In the subroutine SR10, the swallowing simulation device 1 calculates the muscle activity rate α k,0 The new optimized muscle activation rate α kmax,pend Then, the process proceeds to step S64. kmax,pend The analysis process for obtaining this will be described later with reference to FIGS. 22A and 22B.
[0316] In step S64, the swallowing simulation device 1 determines the optimization end time (ts+Δt opt ) muscle activation rate α k,0 The muscle activity rate α obtained in the subroutine SR10 kmax,pend At this time, the swallowing simulation device 1 updates the muscle activity rate α ts and the optimization end time (ts+△t opt ) Muscle activation rate α kmax,pend Linearly interpolate between ts and ts+△t opt The muscle activity rate during the period is also calculated. ts is expressed using the optimization start time ts, but when expressed in terms of time t from the start of the swallowing simulation, the muscle activation rate α t It can also be expressed as:
[0317] In step S65, the swallowing simulation device 1 performs a process of calculating the next optimization target time ts to (ts+Δt opt If a positive result is obtained, this means that the next optimization target time ts~(ts+△t opt ) can be set, that is, dynamic optimization processing has not been performed for all times from the swallowing start time to the swallowing end time of the swallowing simulation. In this case, the swallowing simulation device 1 proceeds to the next step S66.
[0318] In step S66, the swallowing simulation device 1 calculates ts by (ts+Δt opt ), and then (ts+△t opt ) is regarded as ts, and (ts←ts+△t opt), and then the process returns to step S60, and the dynamic optimization process is performed on a new optimization target time between the swallowing start time and the swallowing end time (i.e., as shown in FIG. 21, the new optimization target time is Δt opt Shifted (ts+△t opt )~(ts+2△t opt ) is set, and the above-mentioned processing is repeated until a negative result is obtained in step S65.
[0319] If a negative result is obtained in step S65, this means that the next optimization target time ts~(ts+△t opt ) is not available for setting, that is, the dynamic optimization process has been performed from the swallowing start time to the swallowing end time of the swallowing simulation, and at this time, the swallowing simulation device 1 ends the dynamic optimization process.
[0320] Next, the analysis process of the subroutine SR10 in Fig. 20 will be described below. Fig. 22A and Fig. 22B are flowcharts showing the procedure of the analysis process of the subroutine SR10 in Fig. 20. As shown in Fig. 22A, in step S70, the swallowing simulation device 1 calculates the muscle activity rate α ts The optimization end time (ts+△t opt ) Muscle activation rate α k,0 Linearly interpolate between the optimization target time ts~(ts+△t opt ) is calculated, and the process proceeds to step S71.
[0321] In step S71, the swallowing simulation device 1 calculates the muscle activity rate α k,0(t+△t) is determined, and this is set as the muscle activity rate to be used in one cycle of calculation processing by the particle method, and the process proceeds to the next subroutine SR11. opt △t<△t opt The above relationship exists.
[0322] In the subroutine SR11, at the beginning, the first subroutine SR10, that is, the optimization target time is (swallowing start) ~ (swallowing start + Δt opt ), the position r of muscle particle i at the calculation time (the start of swallowing) finally calculated in the static optimization process of FIG. 17B t+△t and velocity v t+△t and the muscle activity rate α set in step S71. k,0(t+△t) Using the above, in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", one cycle of calculation processing SRb by the particle method shown in FIG. 10 is performed by the motion analysis unit 50, and the process proceeds to the next step S72. opt ) at time (Δt opt )~(2Δt opt ) is the final calculation time Δt calculated in the first subroutine SR10. opt The position r of muscle particle i at kmax,pend and velocity v kmax,pend Similarly, for the third and subsequent times, the position r of muscle particle i finally calculated in the previous subroutine SR10 is used. kmax,pend and velocity v kmax,pend is used.
[0323] As a result, in the swallowing simulation device 1, one cycle of calculation processing SRb by the particle method is performed in the subroutine SR11, so that the position r of the muscle particle i at the calculation time (t+△t) is calculated. t ´ and speed v t ' (step S9 in FIG. 10). Here, the position r of muscle particle i obtained in subroutine SR11 can be calculated. t ´ is r i k,0(t+△t) and the velocity v of muscle particle i calculated in subroutine SR6. t ´ is v i k,0(t+△t) It is written as follows.
[0324] In step S72, the swallowing simulation device 1 determines whether the calculation time (t+Δt) of one cycle of the calculation process SRb using the particle method performed in the subroutine SR11 is equal to or shorter than the optimization end time (ts+Δt opt ) is determined.
[0325] If a negative result is obtained, this means that the calculation time (t+△t) of one cycle of the calculation process SRb by the particle method performed in the subroutine SR11 is equal to or shorter than the optimization end time (ts+△t opt ), and in this case, the swallowing simulation apparatus 1 proceeds to the next step S73.
[0326] In step S73, the swallowing simulation device 1 changes t to t+Δt (t←t+Δt) and proceeds to the next step S74. For example, when the calculation time is the first calculation time (t+Δt), the calculation time is set to (t+2Δt) by changing t at the calculation time (t+Δt) to t+Δt in step S73. In step S74, the swallowing simulation device 1 changes the position r of the muscle particle i to be used in the next subroutine SR11. t and velocity v t (Step S1 in FIG. 10) is calculated by the subroutine SR11. i k,0(t+△t) and velocity v i k,0(t+△t) Then, the process returns to step S71.
[0327] In step S71, the swallowing simulation device 1 calculates the optimum time ts to (ts+Δt opt ), the muscle activity rate α at the calculation time (t+2△t) updated in step S73 k,0(t+2△t) is identified and set as a new muscle activity rate to be used in one cycle of calculation processing using the particle method, and the process proceeds to the next subroutine SR11.
[0328] In the subroutine SR11, the position r of the muscle particle i updated in step S74 is i k,0(t+△t) and velocity v ik,0(t+△t) and the muscle activity rate α newly set in step S71. k,0(t+2△t) Using these, the motion analysis unit 50 again performs one cycle of calculation processing SRb using the particle method shown in Figure 10 in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", and then proceeds to the next step S72.
[0329] Then, the swallowing simulation device 1 continues to perform the process until a positive result is obtained in step S72, that is, until the calculation time (t+Δt) reaches the optimization end time (ts+Δt opt ), the above-mentioned steps S71, subroutine SR11, and steps S72 to S74 are repeated.
[0330] If a positive result is obtained in step S72, this means that the calculation time (t+Δt) is equal to the optimization end time (ts+Δt opt ), that is, the above-mentioned step S71, subroutine SR11, and steps S72 to S74 are repeated, and the optimization end time (ts+Δt opt ) the position r of muscle particle i i k,0(ts+△topt) and velocity v i k,0(ts+△topt) This indicates that the optimization end time (ts+Δt opt ) the position r of muscle particle i i k,0(ts+△topt) is position r i k,0 and the optimization end time (ts+△t opt ) the velocity v of muscle particle i at i k,0(ts+△topt) is the velocity v i k,0 It is written as follows.
[0331] In step S77, the swallowing simulation device 1 determines (i) the optimization end time (ts+Δt opt ) Muscle activation rate α k,0 and (ii) the optimization end time (ts+△topt ) the position r of muscle particle i i k,0 And (iii) the end time of optimization (ts+△t opt ) the velocity v of muscle particle i at i k,0 Using these, the evaluation function e is calculated from the following equation (34). k,0 Then, the process proceeds to step S78.
number
[0332] The above equation (34) used in the dynamic optimization process differs from equation (25) used in the static optimization process in that an arithmetic term related to speed is provided in parentheses in the first term on the right side.
[0333] In the above equation (34), e k,0 indicates the evaluation function used in the dynamic optimization process. Os, Oj, w P,Oj , r i k,0 , s i , w α Since is the same as the above equation (25), the explanation is omitted here. m k,0 is the end time of optimization (ts+△t opt ) Muscle activation rate α k,0 The subscript m indicates the muscle type. i k,0 is the optimization end time (ts+△t opt ) indicates the position of muscle particle i.
[0334] w V,Oj is the velocity v in the head and neck organ (Oj) for which the evaluation function is calculated. i k,0 This constant w V,Oj is determined by experience. For example, w P,Oj ×10 -8 ~w P,Oj ×10 -4 is desirable.
[0335] v i k,0 is the optimization end time (ts+△t opt ) indicates the velocity of muscle particle i. The subscript i is an identifier that identifies muscle particle i. The superscript 0 indicates that muscle particle i is in the initial state where the muscle activity rate value set for the muscle particle for each muscle type has not been changed.
[0336] u i is the target velocity (u i =△s i / △t opt Such a target velocity of the muscle particle i is specified in advance based on anatomical knowledge, VF images when the subject swallows, 4DCT images, and the like, and is stored in the storage unit 83.
[0337] The number of optimization calculations k is determined by the search direction ya k This is the number of times that the final evaluation function is determined by performing the determination of the evaluation function and the line search, and is set in advance by the optimization number setting unit 97 of the swallowing simulation device 1. Here, the maximum number of optimization calculation times k is also k max It is written as follows.
[0338] In addition, the above evaluation function e used in this dynamic optimization process k,0 The general formula is given by the following formula (35).
number
[0339] The superscript Ph indicates an identifier for identifying the change state of the muscle activity rate when calculating the evaluation function while changing the value of the muscle activity rate set for muscle particle i for each specified muscle type. For example, when the muscle activity rate is in the initial value state without being changed, "Ph" is "0", and e k,0 , r i k,0 , vi k,0 , α m k,0 When the value of the muscle activity rate of a specific muscle type m is changed, "Ph" becomes "▽;m" (m indicates the muscle type, 1 to n), and e k,▽;m , r i k,▽;m , v i k,▽;m , α m k,▽;m In addition, the search direction ya k When determining and searching the line, "Ph" becomes "p" and e k,p , r i k,p , v i k,p , α m k,p Furthermore, when the muscle activation rate is finally optimized after the optimization calculation count k, "Ph" becomes "pend" and e k,pend , r i k,pend , v i k,pend , α m k,pend When "k" is the maximum number of times "kmax", e kmax,pend , r i kmax,pend , v i kmax,pend , α m kmax,pend It becomes.
[0340] Here, in the formulas (25) and (26) used in the static optimization process already described, since there is no calculation term related to the velocity shown in the brackets of the first term on the right side in the above formulas (34) and (35), the velocity of the muscle particle i of the head and neck organs and the entire head and neck organs may oscillate, making the analysis unstable. In contrast, in the dynamic optimization process, as shown in the above formulas (34) and (35), by providing a calculation term related to the velocity in the brackets of the first term on the right side, it is possible to stabilize the velocity of the muscle particle i of the head and neck organs and the entire head and neck organs, and perform an analysis of the optimal muscle activity rate.
[0341] In step S77, the evaluation function e k,0After calculating the muscle activity rate α k,0 The muscle activity rate α shown in the above equation (27) is calculated by changing the value of k,▽;m The optimization end time (ts+△t opt ) as the muscle activity rate at the time of the muscle activation, and the process proceeds to the next step S79. k,▽;m Similarly to the static optimization process described above, for example, as shown in the above equation (27), the muscle activity rate α m k,0 The value is changed by adding h to the value. h may be a preset value or may be a value set by the developer each time, but it is preferable that h is a small value such as 0.0001 to 0.01.
[0342] Muscle activity rate α to add h m k,0 For example, muscle activity rate α1 k,0 ,…,α n k,0 Among them, the muscle activity rate α of any one muscle type m m k,0 Add h to m. First, let m = 1 and the muscle activity rate α1 k,0 Muscle activity rate α with h added k,▽;1 (α1 k,0 +h,α2 k,0 ,…,α n k,0 ) is used. k,▽;m The value of is changed by the muscle activity rate changing unit 80.
[0343] In step S79, the swallowing simulation device 1 calculates the muscle activity rate α ts The optimization end time (ts+△t opt ) Muscle activation rate α k,▽;m Linearly interpolate between the optimization target time ts~(ts+△t opt ) is calculated, and the process proceeds to the next step S80.
[0344] In step S80, the swallowing simulation device 1 calculates the optimum time ts to (ts+Δt opt ) and muscle activity rate α at the first calculation time (t+△t) k,▽;m(t+△t) is determined, and set as the muscle activity rate to be used in one cycle of calculation processing by the particle method, and the process proceeds to the next subroutine SR12.
[0345] In the subroutine SR12, at the beginning, the first subroutine SR10, that is, the optimization target time is (swallowing start) ~ (swallowing start + Δt opt ), the position r of muscle particle i at the optimization start time (the start of swallowing) finally calculated in the static optimization process of FIG. 17B t and velocity v t and the muscle activity rate α set in step S80 k,▽;m(t+△t) Using the above, in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic 3D head and neck particle model incorporating only internal forces>", one cycle of calculation processing SRb using the particle method shown in FIG. 10 is performed by the motion analysis unit 50, and the process proceeds to the next step S82. opt )~(2Δt opt ) is the final calculation time Δt calculated in the first subroutine SR10. opt The position r of muscle particle i at kmax,pend and velocity v kmax,pend Similarly, for the third and subsequent times, the position r of muscle particle i finally calculated in the previous subroutine SR10 is used. kmax,pend and velocity v kmax,pend is used.
[0346] As a result, in the swallowing simulation device 1, one cycle of calculation processing SRb using the particle method is performed in the subroutine SR12, thereby obtaining the muscle activity rate α k,▽;m(t+△t) The position r of muscle particle i when t ´ and speed v t ' (step S9 in FIG. 10). Here, the position r of muscle particle i obtained in subroutine SR12 can be calculated. t ´ is r i k,▽;m(t+△t)and the velocity v of muscle particle i calculated in subroutine SR12. t ´ is v i k,▽;m(t+△t) It is written as follows.
[0347] In step S82, the swallowing simulation device 1 determines whether the calculation time (t+Δt) of one cycle of the calculation process SRb using the particle method performed in the subroutine SR12 is equal to or shorter than the optimization end time (ts+Δt opt ) is determined.
[0348] If a negative result is obtained, this means that the calculation time (t+△t) of one cycle of the calculation process SRb by the particle method performed in the subroutine SR12 is equal to or shorter than the optimization end time (ts+△t opt ), and in this case, the swallowing simulation apparatus 1 proceeds to the next step S83.
[0349] In step S83, the swallowing simulation device 1 sets t to t+Δt (t←t+Δt) and proceeds to the next step S84. For example, when the calculation time is the first calculation time (t+Δt), in step S83, t of the calculation time (t+Δt) is set to t+Δt, so that the calculation time is (t+2Δt).
[0350] In step S84, the swallowing simulation device 1 calculates the position r of the muscle particle i to be used in the next subroutine SR12. t and velocity v t (Step S1 in FIG. 10) is performed by subtracting the position r i k,▽;m(t+△t) and velocity v i k,▽;m(t+△t) Then, the process returns to step S80.
[0351] In step S80, the swallowing simulation device 1 calculates the optimum time ts to (ts+Δt opt ), the muscle activity rate α at the next calculation time (t+2△t) updated in step S83 k,▽;m(t+2△t)is identified and set as a new muscle activity rate to be used in one cycle of calculation processing using the particle method, and the process moves to the next subroutine SR12.
[0352] In the subroutine SR12, the position r of the muscle particle i updated in step S84 is i k,▽;m(t+△t) and velocity v i k,▽;m(t+△t) and the muscle activity rate α newly set in step S80 k,▽;m(t+2△t) Using these, the motion analysis unit 50 again performs one cycle of calculation processing SRb using the particle method shown in Figure 10 in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", and proceeds to the next step S82.
[0353] Then, the swallowing simulation device 1 continues to perform the process until a positive result is obtained in step S82, that is, until the calculation time (t+Δt) reaches the optimization end time (ts+Δt opt ), the processes of step S80, subroutine SR12, and steps S83 to S84 described above are repeated.
[0354] If a positive result is obtained in step S82, this means that the calculation time (t+Δt) is equal to the optimization end time (ts+Δt opt ), that is, the above-mentioned step S80, subroutine SR12, and steps S82 to S84 are repeated, and the optimization end time (ts+Δt opt ) the position r of muscle particle i i k,▽;m(ts+△topt) and velocity v i k,▽;m(ts+△topt) This indicates that the optimization end time (ts+Δt opt ) the position r of muscle particle i i k,▽;m(ts+△topt) is position r i k,▽;m and the optimization end time (ts+△t opt ) the velocity v of muscle particle i at i k,▽;m(ts+△topt)is the velocity v i k,▽;m It is written as follows.
[0355] In step S85, the swallowing simulation device 1 determines (i) the optimization end time (ts+Δt opt ) Muscle activation rate α k,▽;m and (ii) the optimization end time (ts+△t opt ) the position r of muscle particle i i k,▽;m And (iii) the end time of optimization (ts+△t opt ) the velocity v of muscle particle i at i k,▽;m Using the above equation (35), the evaluation function e k,▽;m Then, the process proceeds to step S86.
[0356] In step S86, the swallowing simulation device 1 calculates the muscle activity rate α k,▽;m It is determined whether the change in muscle activity rate α has been made for all muscle types m (m=1, 2, ..., n), i.e., whether m=n. If a negative result is obtained in step S86, this means that the muscle activity rate α k,▽;m This indicates that the change has not been made to all muscle types m, and in this case, the process proceeds to the next step S89.
[0357] In step S89, the swallowing simulation device 1 calculates the muscle activity rate α k,▽;m m is changed to m+1, and muscle activity rate α k,▽;m+1 Then, m+1 is regarded as m (m←m+1), and the process proceeds to step S78 again. In step S78, the swallowing simulation device 1, for example, has already determined that the muscle activity rate α k,▽;1 (α1 k,0 +h,α2 k,0 ,…,α n k,0 ) is set, the muscle activity rate α k,▽;2 (α1 k,0 ,α2 k,0 +h,…,α n k,0 ) is muscle activity rate α k,▽;m Then, the process moves to the next subroutine SR12.
[0358] As a result, for example, the muscle activity rate α k,▽;2 Regarding muscle activity rate α k,▽;1 Similarly, the processes of step S80, subroutine SR12, and steps S82 to S84 described above are repeated until a positive result is obtained in step S82.
[0359] If a positive result is obtained in step S86, this means that the muscle activity rate α k,▽;m The change has been made to all muscle types m, that is, the optimization end time (ts + △t opt ) Evaluation function e k,▽;m (e k,▽;1 ,e k,▽;2 ,…,e k,▽;n ) is obtained, and at this time, the swallowing simulation device 1 proceeds to step S40 shown in FIG. 22B.
[0360] Steps S40 to S43 are the same as the static optimization process described above. In step S40, the swallowing simulation device 1 calculates the evaluation function e k,0 , e k,▽;m Using the above equations (28) and (29), the gradient ▽e k Then, the process proceeds to step S41.
[0361] In step S41, the swallowing simulation device 1 calculates a search direction ya using the above formula (30) or (31). k Then, the process moves to step S42, in which a predetermined step size μ p Then, the process proceeds to step S43.
[0362] In step S43, the swallowing simulation device 1 calculates the search direction ya k and the step size μ set in step S42 p and muscle activity rate α k,0Using the above equation (32), the muscle activity rate α k,p Then, the process proceeds to step S90.
[0363] In step S90, the swallowing simulation device 1 calculates the muscle activity rate α k,p The optimization end time (ts+△t opt ) as the muscle activity rate at the optimization start time ts, and the process proceeds to the next step S91. In step S91, the swallowing simulation device 1 sets the muscle activity rate α ts The optimization end time (ts+△t opt ) Muscle activation rate α k,p Linearly interpolate between the optimization target time ts~(ts+△t opt ) is calculated, and the process proceeds to step S92.
[0364] In step S92, the swallowing simulation device 1 calculates the optimum time ts to (ts+Δt opt ) and muscle activity rate α at the first calculation time (t+△t) k,p(t+△t) is determined, and set as the muscle activity rate to be used in one cycle of calculation processing by the particle method, and the process proceeds to the next subroutine SR13.
[0365] In the subroutine SR13, the first subroutine SR10, that is, the optimization target time is (swallowing start) ~ (swallowing start + Δt opt ), the position r of muscle particle i at the optimization start time (the start of swallowing) finally calculated in the static optimization process of FIG. 17B t and velocity v t and the muscle activity rate α set in step S92. k,p(t+△t) Using the above, in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic 3D head and neck particle model incorporating only internal forces>", one cycle of calculation processing SRb using the particle method shown in FIG. 10 is performed by the motion analysis unit 50, and the process proceeds to the next step S93. opt )~(2Δt opt) is the final calculation time Δt calculated in the first subroutine SR10. opt The position r of muscle particle i at kmax,pend and velocity v kmax,pend Similarly, for the third and subsequent times, the position r of muscle particle i finally calculated in the previous subroutine SR10 is used. kmax,pend and velocity v kmax,pend is used.
[0366] As a result, in the swallowing simulation device 1, one cycle of calculation processing SRb using the particle method is performed in the subroutine SR13, thereby obtaining the muscle activity rate α k,p(t+△t) The position r of muscle particle i when t ´ and speed v t ' (step S9 in FIG. 10). Here, the position r of muscle particle i obtained in subroutine SR13 can be calculated. t ´ is r i k,p(t+△t) and the velocity v of muscle particle i calculated in subroutine SR13. t ´ is v i k,p(t+△t) It is written as follows.
[0367] In step S93, the swallowing simulation device 1 determines whether the calculation time (t+Δt) of one cycle of the calculation process SRb using the particle method performed in the subroutine SR13 is equal to or shorter than the optimization end time (ts+Δt opt ) is determined.
[0368] If a negative result is obtained, this means that the calculation time (t+△t) of one cycle of the calculation process SRb by the particle method performed in the subroutine SR13 is equal to or shorter than the optimization end time (ts+△t opt ), and in this case, the swallowing simulation apparatus 1 proceeds to the next step S94.
[0369] In step S94, the swallowing simulation device 1 changes t to t+Δt (t←t+Δt) and proceeds to the next step S95. For example, when the calculation time is the first calculation time (t+Δt), in step S94, t at the calculation time (t+Δt) is changed to t+Δt, so that the calculation time becomes (t+2Δt). In step S95, the swallowing simulation device 1 sets the position r of the muscle particle i to be used in the next subroutine SR13. t and velocity v t (Step S1 in FIG. 10) is performed by subtracting the position r i k,p(t+△t) and velocity v i k,p(t+△t) Then, the process returns to step S92.
[0370] In step S92, the swallowing simulation device 1 calculates the optimum time ts to (ts+Δt opt ), the muscle activity rate α at the calculation time (t+2△t) updated in step S94 k,p(t+2△t) is specified and set as the muscle activity rate to be used in one cycle of calculation processing using the particle method, and the process proceeds to the next subroutine SR13.
[0371] In subroutine SR13, the position r of muscle particle i updated in step S95 is i k,p(t+△t) and velocity v i k,p(t+△t) and the muscle activity rate α set in step S92. k,p(t+2△t) Using these, the motion analysis unit 50 again performs one cycle of calculation processing SRb using the particle method shown in Figure 10 in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", and proceeds to the next step S93.
[0372] Then, the swallowing simulation device 1 continues to perform the process until a positive result is obtained in step S93, that is, until the calculation time (t+Δt) reaches the optimization end time (ts+Δt opt ), the above-mentioned steps S92, subroutine SR13, and steps S93 to S95 are repeated.
[0373] If a positive result is obtained in step S93, this means that the calculation time (t+Δt) is equal to the optimization end time (ts+Δt opt ), that is, the above-mentioned step S92, subroutine SR13, and steps S93 to S95 are repeated, and the optimization end time (ts+Δt opt ) the position r of muscle particle i i k,p(ts+△topt) and velocity v i k,p(ts+△topt) This indicates that the optimization end time (ts+Δt opt ) the position r of muscle particle i i k,p(ts+△topt) is position r i k,p and the optimization end time (ts+△t opt ) the velocity v of muscle particle i at i k,p(ts+△topt) is the velocity v i k,p It is written as follows.
[0374] In step S96, the swallowing simulation device 1 determines (i) the optimization end time (ts+Δt opt ) Muscle activation rate α k,p and (ii) the optimization end time (ts+△t opt ) the position r of muscle particle i i k,p And (iii) the end time of optimization (ts+△t opt ) the velocity v of muscle particle i at i k,p Using the above equation (35), the evaluation function e k,p Then, the process proceeds to step S97.
[0375] In step S97, the swallowing simulation device 1 calculates the evaluation function e k,p As described above, we determine whether or not e in the above formula (33) satisfies the condition. k,(p+1) is the evaluation function ek,p After calculating, the step size μ p Increase the value of step size μ p+1 Specifically, the evaluation function e k,p After calculating, the process proceeds to step S98, where the step size μ p p is set to p+1 and the step size μ p+1 Then, p+1 is regarded as p, and the step size μ p In step S42, the swallowing simulation device 1 calculates the step size μ p Step size μ p+1 Step size μ p In step S43, the muscle activity rate α is calculated from the above equation (32). k,(p+1) Request.
[0376] Thereafter, in the swallowing simulation device 1, the above-mentioned processes of step S92, subroutine SR13, and steps S93 to S95 are repeated until a positive result is obtained in step S93, and the evaluation function e calculated in step S96 is k,(p+1) Using the previously calculated evaluation function e k,p It is judged whether or not satisfies the condition of the above formula (33).
[0377] If a positive result is obtained in step S97, this means that the evaluation function e calculated in step S96 k,p indicates that the condition of the above formula (33) is satisfied, and at this time, the swallowing simulation device 1 proceeds to the next step S99. Note that in step S97, the evaluation function e k,p is the evaluation function finally obtained after the optimization calculation count k. Hereinafter, the evaluation function e k,pend and the evaluation function e k,pend The muscle activity rate at this time is called muscle activity rate α k,pend It is written as follows.
[0378] In step S99, the swallowing simulation device 1 determines whether the optimization calculation count k, which indicates the number of times all of the above-mentioned steps S70 to S97 have been performed, is equal to or greater than the maximum number k preset in the optimization count setting unit 97. max Determine whether it has been reached.
[0379] If a negative result is obtained, this means that the number of optimization calculations k, which is the number of times all of the above-mentioned steps S70 to S97 have been performed, has reached the maximum number k max In this case, the swallowing simulation apparatus 1 proceeds to step S100. In step S100, the swallowing simulation apparatus 1 performs the muscle activity rate α k,pend Set k to k+1 (for example, when the optimization calculation number is k=1 (first time), the next optimization calculation number is k=2 (second time)), set pend to 0, and set the muscle activation rate α k+1,0 Toshi (α k+1,0 ←α k,pend ), and muscle activity rate α k+1,0 Consider k+1 as k and calculate muscle activity rate α k,0 Then, (k←k+1), and the process returns to step S70.
[0380] Specifically, in step S97, the evaluation function e that satisfies the condition of the above formula (33) is k,pend The muscle activity rate α used to calculate k,pend The optimization end time (ts+△t opt ) Muscle activation rate α k,0 In step S70, the swallowing simulation device 1 sets the optimized muscle activity rate α ts and the newly set optimization end time (ts+△t opt ) Muscle activation rate α k,0 Linearly interpolate between the optimization target time ts~(ts+△t opt ) and then proceed to the next step S71.
[0381] After that, the above-mentioned steps S71 to S97 are performed to obtain the muscle activity rate α k,pendThe swallowing simulation device 1 repeats the process of steps S70 to S97 until a positive result is obtained in step S99, that is, until the maximum number of times k max This is repeated until the evaluation function e kmax,pend Then, the evaluation function e kmax,pend Muscle activation rate α kmax,pend Let be the final muscle activation rate of the dynamic optimization process.
[0382] The swallowing simulation device 1 finally calculates the evaluation function e kmax,pend and muscle activity rate α kmax,pend Calculating the optimization target time ts~(ts+△t opt ) analysis process is terminated, and the process proceeds to step S64 shown in FIG.
[0383] (7-3) <Dynamic optimization process of another embodiment performing relaxation processing> Next, a dynamic optimization procedure of another embodiment, which is a technique different from the dynamic optimization procedure shown in FIG. 20 and performs relaxation processing, will be described below.
[0384] FIG. 24 shows the muscle activation rate α finally obtained by the dynamic optimization procedure of another embodiment in which the relaxation process is performed. m t In Figure 24, the solution α1 at time 0.10000S is shown. 0.10000 " and "Solution α2 0.10000 " indicates the muscle activity rate obtained by static optimization processing.
[0385] In the dynamic optimization process of another embodiment in which the relaxation process is performed, the muscle activity rate α m t1 (In Figure 24, "Solution α1 0.10000 " and "Solution α2 0.10000 ") is calculated by static optimization processing, and the muscle activity rate α m t2 ,α m t3 ,…(In FIG. 24, at time t2, “α1 0.10005, α2 0.10005 " At time t3, "α1 0.10010 , α2 0.10010 ") is calculated by dynamic optimization processing, which will be described later. Therefore, the swallowing simulation device 1 does not need to calculate muscle activity rates from all of the multiple frame images by static optimization processing, and the processing load can be reduced accordingly.
[0386] In addition, in the dynamic optimization process of the other embodiment in which the relaxation process is performed, the interval between the optimization target times set when performing the dynamic optimization process is extremely short (Δt opt is 0.0005S), and is characterized in that a relaxation process is performed using a relaxation degree β. In this other embodiment, as shown in FIG. 1, a relaxation process unit 98 is provided in the muscle activity rate analysis unit 91, and the relaxation process is performed by this relaxation process unit 98.
[0387] Here, Fig. 25 is a flowchart showing a dynamic optimization process procedure of another embodiment in which the mitigation process is performed. As shown in Fig. 25, in the swallowing simulation device 1, the process proceeds from the start step to step S110, and as shown in Fig. 26, the dynamic optimization process is performed from the start of swallowing to the end of swallowing in the swallowing simulation at optimization target time ts~(ts+Δt opt ) and proceed to the next step S111. Here, in step S110, Δt opt For example, an extremely small time of 0.00001 to 0.00050 S is set as the time.
[0388] In this case, in the swallowing simulation device 1, for example, the first optimization target time ts~(ts+△t opt ) muscle activity rate α t1 Therefore, in step S111, first, the muscle activity rate α ts The muscle activity rate α at the initial time t1 calculated by the static optimization process is t1 Then, the process proceeds to step S112.
[0389] In step S112, the swallowing simulation device 1 calculates the muscle activity rate α ts (Here, at the initial time t1, the muscle activity rate α t1 (also denoted as ts+△t opt ) muscle activity rate α k,0 Then, the process proceeds to the next subroutine SR15. The analysis process of the subroutine SR15 will be described later with reference to the flowchart of FIG.
[0390] In the subroutine SR15, the swallowing simulation device 1 determines the optimization end time (ts+△t opt ) muscle activation rate α k,0 The optimized muscle activation rate α kmax,pend(β) Then, the process proceeds to step S113. Note that the superscript (β) is an identifier indicating that relaxation processing has been performed.
[0391] In step S113, the swallowing simulation device 1 determines the optimization end time (ts+Δt opt ) muscle activation rate α k,0 The muscle activity rate α obtained in subroutine SR15 kmax,pend(β) Then, the process moves to step S114.
[0392] In step S114, the swallowing simulation device 1 performs a process of calculating the next optimization target time ts to (ts+Δt opt ) can be set.
[0393] If a positive result is obtained, this means that the next optimization target time ts~(ts+△t opt ) can be set, that is, dynamic optimization processing has not been performed for all times from the swallowing start time to the swallowing end time of the swallowing simulation. In this case, the swallowing simulation device 1 proceeds to the next step S115.
[0394] In step S115, the swallowing simulation device 1 calculates ts by (ts+Δt opt ), and then (ts+△t opt ) is regarded as ts, and (ts←ts+△t opt ), and then the process returns to step S110, and the next optimization target time for which dynamic optimization processing is performed is selected between the swallowing start time and the swallowing end time of the swallowing simulation (i.e., in FIG. 26, the new optimization target time is Δt opt Shifted (ts+△t opt )~(ts+2△t opt ) is set, and the above-mentioned processing is repeated until a negative result is obtained in step S114.
[0395] If a negative result is obtained in step S114, this means that the next optimization target time ts~(ts+Δt opt ) is not available for setting, that is, the dynamic optimization process has been performed from the swallowing start time to the swallowing end time of the swallowing simulation, and at this time, the swallowing simulation device 1 ends the dynamic optimization process.
[0396] Next, a description will be given of the analysis process of the subroutine SR15 in Fig. 25. Fig. 27 is a flowchart showing the procedure of the analysis process of the subroutine SR15 in Fig. 25. As shown in Fig. 27, the swallowing simulation device 1 moves from the start step to the subroutine SR16.
[0397] In the subroutine SR16, at the first subroutine SR15 (FIG. 25), that is, when the optimization target time is (start of swallowing) to (start of swallowing + Δt opt ), the position r of muscle particle i at the optimization start time (the start of swallowing) finally calculated in the static optimization process of FIG. 17B t and velocity v t and the optimization end time (ts+△t opt ) muscle activity rate α k,0Using these, in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", one cycle of calculation processing SRb using the particle method shown in Figure 10 is performed by the motion analysis unit 50, and then the process proceeds to the next step S120.
[0398] Here, if the first analysis process of the subroutine SR15 has already been performed, the subroutine SR16 performed in the second analysis process of the subroutine SR15 is as follows. In this case, the second optimization target time is (ts + △t opt )~(ts+2△t opt ) Here, the second optimization start time (ts+△t opt ) is simply denoted as t2, and the end time of the second optimization (ts+2△t opt ) simply as (t2+Δt opt ) Note that the optimization start time (swallowing start) of the first subroutine SR15 is t1, and the optimization end time of the first subroutine SR15 is (t1 + Δt opt ), the optimization start time t2 of the second subroutine SR15 is the optimization end time (t1 + Δt opt ) and t2=(t1+Δt opt ) relationship.
[0399] At the second subroutine SR15 (Fig. 25), that is, the second optimization target time (t2) to (t2+Δt opt ), in the first analysis of subroutine SR15, the muscle activity rate α kmax,pend(β) The end time of the first optimization of subroutine SR15 (t1+△t opt )((t1+Δt opt )=t2) kmax,pend(β) and velocity v kmax,pend(β) and the muscle activity rate α set in step S112. k,0Using these, one cycle of calculation processing SRb is performed using the particle method shown in Figure 10 according to the above-mentioned "(5-2) <Structural analysis method for dynamic 3D head and neck particle model incorporating only internal forces>". Note that here, the position of muscle particle i obtained in subroutine SR20, which will be described later, is calculated by r kmax,pend(β) and the velocity of muscle particle i calculated in subroutine SR20 is v kmax,pend(β) It is written as follows.
[0400] That is, in the subroutine SR16 executed in the second or subsequent subroutine SR15, the α obtained in the immediately preceding subroutine SR15 is kmax,pend(β) The position r of muscle particle i obtained by performing one cycle of calculation processing using the particle method in the immediately preceding subroutine SR15 (subroutine SR20 described later) using kmax,pend(β) and velocity v kmax,pend(β) and the position r of muscle particle i in this subroutine SR16. t and velocity v t is used as the initial value for one cycle of calculation processing SRb using the particle method shown in FIG.
[0401] As a result, in the swallowing simulation device 1, one cycle of calculation processing SRb using the particle method is performed in the subroutine SR16, and the optimization end time (ts+△t opt ) the position r of muscle particle i t ´ and speed v t ' (Step S9 in FIG. 10). Here, the position r of muscle particle i obtained in subroutine SR16 can be calculated. t ´ is r i k,0 and the velocity v of muscle particle i calculated in subroutine SR16. t ´ is v i k,0 It is written as follows.
[0402] In step S120, the swallowing simulation device 1 determines (i) the optimization end time (ts+Δt opt ) muscle activity rate α k,0 and (ii) the position r of muscle particle i calculated in subroutine SR16. ik,0 and (iii) the velocity v of muscle particle i, calculated in subroutine SR16. i k,0 Using the above equation (34), the evaluation function e k,0 Then, the process proceeds to step S121.
[0403] In step S121, the swallowing simulation device 1 calculates the muscle activity rate α k,0 The muscle activity rate α shown in the above equation (27) is calculated by changing the value of k,▽;m The optimization end time (ts+△t opt ) as the muscle activity rate, and the process moves to the next subroutine SR17. k,▽;m Similarly to the static optimization process described above, for example, as shown in the above equation (27), the muscle activity rate α m k,0 is changed by adding h to it. h may be a preset value or may be a value set by the developer each time, but it is preferable that h is a small value such as 0.0001 to 0.01.
[0404] Muscle activity rate α to add h m k,0 For example, muscle activity rate α1 k,0 ,…,α n k,0 Among them, the muscle activity rate α of any one muscle type m m k,0 Add h to m. First, let m = 1 and the muscle activity rate α1 k,0 Muscle activity rate α with h added k,▽;1 (α1 k,0 +h,α2 k,0 ,…,α n k,0 ) is used. k,▽;m The value of is changed by the muscle activity rate changing unit 80.
[0405] In the subroutine SR17, the first time of the subroutine SR15, that is, the optimization target time is (swallowing start) ~ (swallowing start + Δt opt), the position r of muscle particle i at the optimization start time (the start of swallowing) finally calculated in the static optimization process of FIG. 17B t and velocity v t and the optimization end time (ts+△t opt ) Muscle activation rate α k,▽;m Using these, in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", one cycle of calculation processing SRb using the particle method shown in Figure 10 is performed by the motion analysis unit 50, and the process proceeds to the next step S122.
[0406] As described above, at the second subroutine SR15 (FIG. 25), that is, at the second optimization target time (t2) to (t2+Δt opt ) In the subroutine SR17, the muscle activity rate α kmax,pend(β) The end time of the first optimization of subroutine SR15 (t1+△t opt ) the position r of muscle particle i kmax,pend(β) and velocity v kmax,pend(β) and the muscle activity rate α newly set in step S121. k,▽;m Using these, one cycle of calculation processing SRb is performed using the particle method shown in Figure 10 in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic 3D head and neck particle model incorporating only internal forces>".
[0407] That is, in the subroutine SR17 executed in the second or subsequent subroutine SR15, the α obtained in the immediately preceding subroutine SR15 is kmax,pend(β) The position r of muscle particle i obtained by performing one cycle of calculation processing using the particle method in the immediately preceding subroutine SR15 (subroutine SR20 described later) using kmax,pend(β) and velocity v kmax,pend(β) and the position r of muscle particle i in this subroutine SR17. t and velocity v t is used as the initial value for one cycle of calculation processing SRb using the particle method shown in FIG.
[0408] As a result, in the swallowing simulation device 1, one cycle of calculation processing SRb using the particle method is performed in the subroutine SR17, and the optimization end time (ts+△t opt ) the position r of muscle particle i t ´ and speed v t ' (step S9 in FIG. 10). Here, the position r of muscle particle i obtained in subroutine SR17 can be calculated. t ´ is r i k,▽;m and the velocity v of muscle particle i calculated in subroutine SR17. t ´ is v i k,▽;m It is written as follows.
[0409] In step S122, the swallowing simulation device 1 determines (i) the optimization end time (ts+Δt opt ) Muscle activation rate α k,▽;m and (ii) the position r of muscle particle i calculated by subroutine SR17. i k,▽;m and (iii) the velocity v of muscle particle i, calculated in subroutine SR17. i k,▽;m Using the above equation (35), the evaluation function e k,▽;m Then, the process proceeds to step S123.
[0410] In step S123, the swallowing simulation device 1 calculates the muscle activity rate α k,▽;m It is determined whether the change in muscle activity rate α in step S121 has been performed for all muscle types m (m=1, 2, ..., n), i.e., whether m=n. If a negative result is obtained in step S123, this means that the change in muscle activity rate α in step S121 has been performed for all muscle types m (m=1, 2, ..., n). k,▽;m This indicates that the change has not been made to all muscle types m, and in this case, the process proceeds to the next step S124.
[0411] In step S124, the swallowing simulation device 1 calculates the muscle activity rate α k,▽;m In step S121, the muscle activity rate α k,▽;1 (α1k,0 +h,α2 k,0 ,…,α n k,0 ) is set, in the second step S121, the muscle activity rate α k,▽;2 (α1 k,0 ,α2 k,0 +h,…,α n k,0 ) and proceeds to the next subroutine SR17.
[0412] As a result, for example, the muscle activity rate α k,▽;2 Regarding muscle activity rate α k,▽;1 Similarly, the above-mentioned subroutine SR17 and step S122 are executed.
[0413] If a positive result is obtained in step S123, this means that the muscle activity rate α k,▽;m The change in evaluation function e has been made for all muscle types m, that is, in step S122, k,▽;m (e k,▽;1 ,e k,▽;2 ,…,e k,▽;n ) is obtained, and at this time, the swallowing simulation device 1 proceeds to the next step S40.
[0414] Steps S40 to S43 are the same as the static optimization process described above. In step S40, the swallowing simulation device 1 calculates the evaluation function e k,0 , e k,▽;m Using the above equations (28) and (29), the gradient ▽e k Then, the process proceeds to step S41.
[0415] In step S41, the swallowing simulation device 1 calculates a search direction ya using the above formula (30) or (31). k Then, the process moves to step S42, in which a predetermined step size μ p Then, the process proceeds to step S43.
[0416] In step S43, the swallowing simulation device 1 calculates the search direction ya k and the step size μ set in step S42 p and muscle activity rate α k,0 Using the above equation (32), the muscle activity rate α k,p Then, the process proceeds to step S126.
[0417] In step S126, the swallowing simulation device 1 calculates the muscle activity rate α k,p The optimization end time (ts+△t opt ) Muscle activation rate α k,p and proceeds to the next subroutine SR18.
[0418] In the subroutine SR18, the first time of the subroutine SR15, that is, the optimization target time is (swallowing start) ~ (swallowing start + Δt opt ), the position r of muscle particle i at the optimization start time (the start of swallowing) finally calculated in the static optimization process of FIG. 17B t and velocity v t and the optimization end time (ts+△t opt ) muscle activity rate α k,p Using these, in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", one cycle of calculation processing SRb using the particle method shown in Figure 10 is performed by the motion analysis unit 50, and the process proceeds to the next step S127.
[0419] As described above, at the second subroutine SR15 (FIG. 25), that is, at the second optimization target time (t2) to (t2+Δt opt ) In the subroutine SR18, the muscle activity rate α kmax,pend(β) The end time of the first optimization of the subroutine SR15 (t1+△t opt ) the position r of muscle particle i kmax,pend(β) and velocity v kmax,pend(β) and the muscle activity rate α newly set in step S43. k,pUsing these, one cycle of calculation processing SRb is performed using the particle method shown in Figure 10 in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic 3D head and neck particle model incorporating only internal forces>".
[0420] That is, in the subroutine SR18 executed in the second or subsequent subroutine SR15, the α obtained in the immediately preceding subroutine SR15 is kmax,pend(β) The position r of muscle particle i obtained by performing one cycle of calculation processing using the particle method in the immediately preceding subroutine SR15 (subroutine SR20 described later) using kmax,pend(β) and velocity v kmax,pend(β) and the position r of muscle particle i in this subroutine SR18. t and velocity v t is used as the initial value for one cycle of calculation processing SRb using the particle method shown in FIG.
[0421] As a result, in the swallowing simulation device 1, one cycle of calculation processing SRb using the particle method is performed in the subroutine SR18, and the optimization end time (ts+△t opt ) the position r of muscle particle i t ´ and speed v t ' (step S9 in FIG. 10). Here, the position r of muscle particle i obtained in subroutine SR18 can be calculated. t ´ is r i k,p and the velocity v of muscle particle i calculated in subroutine SR17. t ´ is v i k,p It is written as follows.
[0422] In step S127, the swallowing simulation device 1 determines (i) the optimization end time (ts+Δt opt ) muscle activity rate α k,p and (ii) the position r of muscle particle i calculated by subroutine SR18. i k,p and (iii) the velocity v of muscle particle i, calculated by subroutine SR18. i k,p Using the above equation (35), the evaluation function ek,p Then, the process proceeds to step S128.
[0423] In step S128, the swallowing simulation device 1 calculates the evaluation function e k,p As described above, we determine whether or not e in the above formula (33) satisfies the condition. k,(p+1) is the evaluation function e k,p After calculating, the step size μ p Increase the value of step size μ p+1 Specifically, the evaluation function e k,p After calculating, the process proceeds to step S129, where p is set to p+1 (p←p+1), and the process returns to step S42. In step S42, the swallowing simulation device 1 calculates the step size μ p Increase the value of step size μ p+1 Then, in step S43, the muscle activity rate α k,(p+1) Request.
[0424] Thereafter, in the swallowing simulation device 1, the processes of the above-mentioned steps S129, S42, S43, S126, subroutine SR18, S127, and S128 are repeated until a positive result is obtained in step S128, and the evaluation function e calculated in step S127 is k,(p+1) Using the previously calculated evaluation function e k,p It is judged whether or not satisfies the condition of the above formula (33).
[0425] If a positive result is obtained in step S128, this means that the evaluation function e calculated in step S127 k,p indicates that the condition of the above formula (33) is satisfied, and at this time, the swallowing simulation device 1 proceeds to the next step S130. Note that in step S128, the evaluation function e k,p is the evaluation function finally obtained after the optimization calculation count k. Hereinafter, the evaluation function e k,pendand the evaluation function e k,pend The muscle activity rate at this time is called muscle activity rate α k,pend It is written as follows.
[0426] In step S130, the swallowing simulation device 1 determines whether the optimization calculation count k, which indicates the number of times all of the processes from the subroutine SR16 to step S128 described above, has been performed, is equal to or greater than the maximum number k preset in the optimization count setting unit 97. max Determine whether it has been reached.
[0427] If a negative result is obtained, this means that the optimization calculation count k, which is the number of times all of the above-mentioned subroutines SR16 to step S128 have been performed, has reached the maximum count k. max In this case, the swallowing simulation apparatus 1 proceeds to step S131. In step S131, the swallowing simulation apparatus 1 performs the processes of the subroutine SR16 to step S128 described above, and the muscle activity rate α k,pend Set k to k+1 (for example, when the optimization calculation number is k=1 (first time), the next optimization calculation number is k=2 (second time)), set pend to 0, and set the muscle activation rate α k+1,0 Toshi (α k+1,0 ←α k,pend ), and muscle activity rate α k+1,0 Consider k+1 as k and calculate muscle activity rate α k,0 Then (k←k+1), and the process returns to subroutine SR16.
[0428] Specifically, in step S128, the evaluation function e that satisfies the condition of the above formula (33) is k,pend The muscle activity rate α used to calculate k,pend The optimization end time (ts+△t opt ) Muscle activation rate α k,0 Set as.
[0429] In the subroutine SR16 transferred from step S131, the first subroutine SR15, that is, the optimization target time is (swallowing start) to (swallowing start + Δt opt), the position r of muscle particle i at the final calculation time (the start of swallowing) calculated by the static optimization process in FIG. 17B t+△t and velocity v t+△t and the muscle activity rate α newly set in step S131. k,0 Using these, in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", one cycle of calculation processing SRb using the particle method shown in Figure 10 is performed by the motion analysis unit 50, and then the process proceeds to the next step S120.
[0430] As described above, in the subroutine SR16 executed in the second or subsequent subroutine SR15, the α obtained in the immediately preceding subroutine SR15 is used. kmax,pend(β) The position r of muscle particle i in the previous subroutine SR15 is obtained by performing one cycle of calculation processing using the particle method (subroutine SR20 described later). kmax,pend(β) and velocity v kmax,pend(β) and the position r of muscle particle i in this subroutine SR16. t and velocity v t is used as the initial value for one cycle of calculation processing SRb using the particle method shown in FIG.
[0431] After that, the above-mentioned steps S120 to S130 are performed, and the muscle activity rate α k,pend The swallowing simulation device 1 performs such a process until a positive result is obtained in step S130, that is, until the maximum number of times k max The evaluation function obtained in the end is the evaluation function e kmax,pend Then, the evaluation function e kmax,pend The muscle activity rate at this time is muscle activity rate α kmax,pend Let us assume that.
[0432] In step S132, the swallowing simulation device 1 performs a relaxation process using the relaxation degree β based on the following formula (36) to obtain a new muscle activity rate α kmax,pend(β) Then, the process moves to the next subroutine SR20. Here, the muscle activity rate α kmax,pend(β)The superscript (β) in is an identifier indicating that the relaxation process has been performed. After that, in step S113 of FIG. 25 described above, the optimization end time (ts+Δt opt ) muscle activity rate α k,0 The muscle activity rate α calculated in step S132 kmax,pend(β) to update.
number
[0433] The relaxation degree β is a constant that is set in advance. The relaxation degree β is preferably greater than 1 and less than 10,000, but may be a value of 10,000 or more. ts is the muscle activity rate at the optimization start time ts, which was specified in step S111 of FIG. 25. Note that the muscle activity rate α ts If the optimization start time ts is time t from the start of the swallowing simulation, the muscle activation rate α t It can also be written as:
[0434] In the subroutine SR20, the first time the subroutine SR15 is executed, that is, the optimization target time is (start of swallowing) ~ (start of swallowing + Δt opt ), the position r of muscle particle i at the optimization start time (the start of swallowing) finally calculated in the static optimization process of FIG. 17B t and velocity v t and the muscle activity rate α obtained in step S132. kmax,pend(β) Using these, in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating only internal forces>", one cycle of calculation processing SRb using the particle method shown in Figure 10 is performed by the motion analysis unit 50, and the above-mentioned analysis processing is terminated.
[0435] Here, at the second subroutine SR15 (Fig. 25), that is, the second optimization target time (t2) to (t2+Δt opt ) in the subroutine SR20, the muscle activity rate αkmax,pend(β) The position r of muscle particle i in the first subroutine SR15 is calculated using kmax,pend(β) and velocity v kmax,pend(β) and the muscle activity rate α newly set in step S132. kmax,pend(β) Using these, one cycle of calculation processing SRb is performed using the particle method shown in Figure 10 in accordance with the above-mentioned "(5-2) <Structural analysis method for dynamic 3D head and neck particle model incorporating only internal forces>".
[0436] That is, in the subroutine SR20 executed in the second or subsequent subroutine SR15, the muscle activity rate α kmax,pend(β) The position r of muscle particle i obtained by performing one cycle of calculation processing using the particle method in the immediately preceding subroutine SR15 (subroutine SR20 described later) using kmax,pend(β) and velocity v kmax,pend(β) and the position r of muscle particle i in this subroutine SR20. t and velocity v t is used as the initial value for one cycle of calculation processing SRb using the particle method shown in FIG.
[0437] As a result, in the swallowing simulation device 1, one cycle of calculation processing SRb using the particle method is performed in the subroutine SR20, so that the muscle activity rate α kmax,pend(β) Using this, the optimization end time (ts+△t opt ) the position r of muscle particle i t ´ and speed v t ' (Step S9 in FIG. 10). Here, the position r of muscle particle i obtained in subroutine SR20 can be calculated. t ´ is r kmax,pend(β) and the velocity v of muscle particle i calculated in subroutine SR20. t ´ is v kmax,pend(β) It is written as follows.
[0438] Then, in subroutine SR20, muscle activity rate α kmax,pend(β) The optimization end time (ts+△t opt ) the position r of muscle particle i kmax,pend(β) and velocity v kmax,pend(β)are set as initial values in the analysis process of the next subroutine SR15.
[0439] Here, a verification test was conducted to examine the change over time of the muscle activity rate during the swallowing simulation by changing the value of the relaxation degree β. As a result, the results shown in Fig. 28A to Fig. 28E were obtained. Here, the dynamic optimization process shown in Fig. 25 was performed using the dynamic three-dimensional head and neck particle model 50a shown in Fig. 11. Then, the relaxation degree β was changed from 1, 10, 100, 1000, to 10000, and the change in the muscle activity rate (here, simply referred to as muscle activity rate α) was confirmed.
[0440] Here, research using electromyography has shown that the temporal change in muscle activity rate is relatively smooth (Reference: "Electromyography of Swallowing with Fine Wire Intramuscular Electrodes in Healthy Human: Activation Sequence of Selected Hyoid Muscles" https: / / www.semanticscholar.org / paper / Electromyography-of-Swallowing-with-Fine-Wire-in-of-Inokuchi-Gonz%C3%A1lez-Fern%C3%A1ndez / e443e2f91d0a9473ed619e2f4b080d8eac2ca09e).
[0441] As shown in FIG. 28A, when the relaxation degree β was set to 1 in the above equation (36), the muscle activity rate α fluctuated finely during the swallowing simulation, and it was confirmed that the temporal change in the muscle activity rate α needs to be smoothed.
[0442] 28B to 28E, it was confirmed that the small oscillations of the muscle activity rate α were suppressed by increasing the relaxation degree β. However, when the relaxation degree β=10000, the waveform of the muscle activity rate α became slightly different from the initial waveform. Therefore, in this embodiment, it was confirmed that the relaxation degree β is preferably more than 1 and less than 10000, and particularly preferably about 10 to 1000.
[0443] Next, the behavior of the hyoid bone 52 of the dynamic three-dimensional head and neck particle model 50a during the swallowing simulation was examined while changing the relaxation degree β from 1, 10, 100, 1000, and 10000. The results shown in Figures 29A to 29E were obtained. Note that the target values in Figures 29A to 29E are obtained by tracing the hyoid bone of the subject based on the VF image, 4DCT image, etc., when the subject swallows.
[0444] 29A to 29D, it was confirmed that the behavior of the hyoid bone 52 was close to the target value when the relaxation degree β was set to 1, 10, 100, and 1000. However, as shown in Fig. 29E, it was confirmed that when the relaxation degree β was set to 10000, the behavior of the hyoid bone 52 deviated relatively significantly from the target value, and from this point of view, it was also confirmed that the relaxation degree β is preferably greater than 1 and less than 10000.
[0445] In the above configuration, the muscle activity rate analysis unit 91 analyzes the movement of the head and neck organs in a three-dimensional image based on the particle method using the dynamic three-dimensional head and neck particle model 50a. The evaluation function e k,p Using the muscle activity rate α kmax,pend Identify the optimization target time ts~(ts+△t opt ) and set the optimization target time ts~(ts+△t opt ) optimization end time (ts+△t opt ) Muscle activation rate α kmax,pend The evaluation function e k,p It was decided to identify them based on the above.
[0446] Then, the muscle activity rate analysis unit 91 calculates the optimization target time ts~(ts+△t opt) muscle activation rate α at optimization start time ts ts (If the optimization start time ts is the time t from the start of the swallowing simulation, the muscle activity rate α t ) and the optimization end time (ts+△t opt ) Muscle activation rate α kmax,pend Using the relaxation degree β, the muscle activity rate α after relaxation is calculated from the above equation (36). kmax,pend(β) In this embodiment, the muscle activity rate α kmax,pend(β) This makes it possible to smooth out the temporal changes in the movement of the head and neck organs during swallowing, and thus makes it possible to reproduce the movement of the head and neck organs during swallowing more accurately than ever before.
[0447] (8) <Other verification results> (8-1) <Calculation of evaluation function and muscle activity rate in manual setting process and static optimization process> Next, manual setting processing and static optimization processing were actually performed using the simplified dynamic three-dimensional head and neck particle model 50a shown in FIG. 11. In the verification test described below, calculation processing (i.e., swallowing simulation) was performed using the particle method according to the above-mentioned "(5-2) <Structural analysis method of dynamic three-dimensional head and neck particle model incorporating only internal forces>". Here, from among the frame images constituting the moving image obtained as the analysis result of the swallowing simulation, the 39th frame image from the start of the swallowing simulation was selected, and manual setting processing and static optimization processing were performed on this frame image. As a result, the results shown in FIG. 30 were obtained.
[0448] In FIG. 30, “ite:k” indicates the number of optimization calculations k, and “ite” of “0” indicates an item of manual setting processing (i.e., the evaluation function e 0,0 indicates the evaluation function in the manual setting process). In the static optimization process, the maximum number of optimization calculations k is maxThere are four optimization calculations, and "ite" has the following items: "1" at the start of the first optimization calculation, "2" at the start of the second optimization calculation, "3" at the start of the third optimization calculation, "4" at the start of the fourth optimization calculation, and "end" after the fourth optimization calculation ends.
[0449] Evaluation function e of "end" after the fourth optimization calculation k,0 The "0.003704" in kmax,pend In the 39th frame image, this evaluation function e kmax,pend The muscle activity rate α for each muscle type when the above was obtained was as shown in Figure 31, and it was confirmed that the muscle activity rate α was a relatively small value.
[0450] (8-2) <Behavior of the hyoid bone in manual setting and static optimization processes> Next, as shown in Fig. 32, a dynamic three-dimensional head and neck particle model 50c in which the left and right stylohyoid muscles 56a, 56b behave separately, and a dynamic three-dimensional head and neck particle model 50d in which the left and right stylohyoid muscles 56a, 56b behave integrally were prepared, and the muscle activity rate α was determined by performing manual setting processing and static optimization processing in the same manner as in the above-mentioned verification test. Then, the difference in the behavior of the hyoid bone 52 of the dynamic three-dimensional head and neck particle models 50c, 50d at that time was confirmed, and the result shown in Fig. 33 was obtained.
[0451] In Figure 33, manual setting processing and static optimization processing were performed for every 9 frames of the moving images obtained during the swallowing simulation to determine the muscle activity rate α of the stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58, and a swallowing simulation was performed using dynamic three-dimensional head and neck particle models 50c, 50d set with the finally determined muscle activity rate α, and the coordinate of the center of gravity of the hyoid bone 52 in the three-dimensional image was confirmed.
[0452] In Fig. 33, the results obtained from dynamic three-dimensional head and neck particle model 50c in which left and right stylohyoid muscles 56a, 56b behave separately are indicated by "+", and the results obtained from dynamic three-dimensional head and neck particle model 50d in which left and right stylohyoid muscles 56a, 56b behave integrally are indicated by "x". Also, in Fig. 33, the position of the hyoid bone identified based on VF images and 4DCT images obtained when the subject swallowed is indicated as a target value by "●".
[0453] 33, it was confirmed that the dynamic three-dimensional head and neck particle model 50c, in which the left and right stylohyoid muscles 56a, 56b behave separately, and the dynamic three-dimensional head and neck particle model 50d, in which the left and right stylohyoid muscles 56a, 56b behave together, exhibit behavior close to the target value. It was also confirmed that the dynamic three-dimensional head and neck particle model 50c, in which the left and right stylohyoid muscles 56a, 56b behave separately, exhibits behavior closer to the target value than the dynamic three-dimensional head and neck particle model 50d, in which the left and right stylohyoid muscles 56a, 56b behave together.
[0454] FIG. 34 summarizes the results obtained from the above-mentioned verification test in which manual setting and static optimization processes were performed every 9 frames, and shows the relationship between the deviation of the center of gravity of the hyoid bone 52, time (frame number), and displacement from the reference position.
[0455] 34, "Result 1: Deviation of center of gravity" indicates the deviation from the target position of the center of gravity of the hyoid bone 52 in dynamic three-dimensional head and neck particle model 50c, in which the left and right stylohyoid muscles 56a, 56b behave separately. "Result 2: Deviation of center of gravity" indicates the deviation from the target position of the center of gravity of the hyoid bone 52 in dynamic three-dimensional head and neck particle model 50d, in which the left and right stylohyoid muscles 56a, 56b behave integrally.
[0456] "Hyoid bone displacement from the reference position on CT" refers to the displacement of the hyoid bone from the "reference position" during swallowing, based on VF images and 4DCT images obtained when the subject swallows, by identifying the position of the hyoid bone before the start of swallowing as the "reference position" and the position to which the hyoid bone moves during swallowing.
[0457] From Figure 34, it can be seen that the dynamic three-dimensional head and neck particle model 50c, in which the left and right stylohyoid muscles 56a, 56b behave separately, has a relatively smaller shift in the center of gravity of the hyoid bone 52 than the dynamic three-dimensional head and neck particle model 50d, in which the left and right stylohyoid muscles 56a, 56b behave integrally.
[0458] Moreover, Fig. 35 summarizes the results obtained from the above-mentioned verification test in which manual setting processing and static optimization processing were performed every 9 frames, and shows the relationship between the evaluation function e, time (number of frames), and the displacement from the reference configuration. From Fig. 35, it can be inferred that there is no particular correlation between the "evaluation function e" and the "displacement from the reference configuration."
[0459] Next, in the above-mentioned verification test in which manual setting processing and static optimization processing were performed every 9 frames, the verification results obtained from the dynamic 3D head and neck particle model 50c in which the left and right stylohyoid muscles 56a, 56b are made to behave separately are shown in Fig. 36. Fig. 36 summarizes the muscle activity rate α for each muscle type, the time (number of frames), and the displacement of the hyoid bone 52 for the optimization result 1.
[0460] Moreover, in the above-mentioned verification test in which manual setting processing and static optimization processing were performed every 9 frames, the verification results (hereinafter also referred to as optimization result 2) obtained from the dynamic three-dimensional head and neck particle model 50d in which the left and right stylohyoid muscles 56a, 56b behave in an integrated manner are shown in Fig. 37. Fig. 37 summarizes the muscle activity rate α for each muscle type, the time (number of frames), and the displacement of the hyoid bone 52 for optimization result 2.
[0461] In Figures 36 and 37, "GenHyo" indicates the "geniohyoid muscle 57," "StyHyoL" indicates the "left stylohyoid muscle 56b," "StyHyoR" indicates the "right stylohyoid muscle 56a," and "SteThy" indicates the "sternohyoid muscle 58."
[0462] 36 and 37, it was confirmed that almost the same results were obtained for optimization result 1 and optimization result 2. In addition, it was confirmed that the muscle activity rate α of each of the stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58 had different timing of change during the swallowing simulation, and the behavior of each muscle activity rate α was important.
[0463] (8-3)<Manual setting and static optimization processing using a dynamic 3D head and neck particle model that faithfully reproduces the head and neck organs> Next, as shown in Fig. 38, a dynamic three-dimensional head and neck particle model 10g that faithfully reproduces the head and neck organs was created, and manual setting processing and static optimization processing were performed. Note that "10h" in Fig. 34 is a cross-sectional view showing the cross-sectional configuration of the dynamic three-dimensional head and neck particle model 10g in the median plane. The dynamic three-dimensional head and neck particle models 10g and 10h were created according to the above-mentioned "(4) <Create a dynamic three-dimensional head and neck particle model>".
[0464] Here, Figure 39A shows the analysis results when the muscle activity rate α was set for each of seven muscle types, namely, "posterior end of genioglossus muscle (GenGlo6 / 6) 61," "hyoglossus muscle (HyoGlo)," "anterior belly of digastric muscle (AntDig) 63," "geniohyoid muscle (GenHyo) 57," "mylohyoid muscle (MylHyo) 64," "posterior belly of digastric muscle (PosDig) 65," and "stylohyoid muscle (StyHyo) 56," by manual setting processing for the frame image of the 57th frame from the start of swallowing.
[0465] The left side of FIG. 39A, "10j1", shows an overall image of the dynamic three-dimensional head and neck particle model in the 57th frame, and the right side, "10J11", shows a midline cross section of the dynamic three-dimensional head and neck particle model 10j1 near the hyoid bone.
[0466] Next, for the frame image of the 57th frame from the start of swallowing, the muscle activity rate α was set for each of the seven muscle types in the same manner as above using a manual setting process, and then the muscle activity rate α for each of the seven muscle types was corrected in the same manner as above using a static optimization process. Figure 39B shows the analysis results.
[0467] The left side of FIG. 39B, "10j2", shows an overall image of the dynamic three-dimensional head and neck particle model in the 674th frame, and the right side, "10J21", shows a midline cross section near the hyoid bone of the dynamic three-dimensional head and neck particle model 10j2.
[0468] FIG. 40 shows the evaluation function e when the manual setting process and the static optimization process are performed on the 57th frame image from the start of swallowing. k,0 In FIG. 40, "ite:k" indicates the number of optimization calculations k, and "ite:k" of "0" indicates an item of manual setting processing (i.e., the evaluation function e 0,0 indicates the evaluation function in the manual setting process. Here, the maximum number of optimization calculations k in the static optimization process is k max The optimization calculation is performed four times, and "ite:k" has the items "1" at the start of the first optimization calculation, "2" at the start of the second optimization calculation, "3" at the start of the third optimization calculation, "4" at the start of the fourth optimization calculation, and "end" at the end of the fourth optimization calculation. kmax,pend This shows that.
[0469] From Figures 39A, 39B, and 40, it was confirmed that even in the dynamic three-dimensional head and neck particle model 10g, which faithfully reproduces the head and neck organs, the muscle activity rate α can be set for each muscle type by manual setting processing and static optimization processing.
[0470] (8-4)<Dynamic optimization process using a dynamic 3D particle model of the head and neck that faithfully reproduces the head and neck organs> Next, a dynamic three-dimensional head and neck particle model that faithfully reproduces the head and neck organs was created, and after performing manual setting processing and static optimization processing, dynamic optimization processing was further performed. Figures 41 and 42 show the analysis results when performing manual setting processing, static optimization processing, and dynamic optimization processing using a dynamic three-dimensional head and neck particle model that faithfully reproduces the head and neck organs, determining the muscle activity rate α for each of the seven muscle types, and performing a swallowing simulation using the dynamic three-dimensional head and neck particle model with the muscle activity rate α set.
[0471] The seven muscle types for which the optimal muscle activity rate α is determined by the manual setting process, static optimization process, and dynamic optimization process are the same as those described above in “(8-3) <Manual setting process and static optimization process using a dynamic three-dimensional head and neck particle model that faithfully reproduces the head and neck organs>”.
[0472] In addition, in this verification test, the dynamic optimization process described in the above "(7-3) <Dynamic optimization process of other embodiments performing mitigation process>" was performed as the dynamic optimization process, and the mitigation degree β at that time was set to 10.
[0473] Figures 41 and 42 show dynamic three-dimensional head and neck particle models 10k1, 10k2, 10k3, and 10k4 of the overall image, dynamic three-dimensional head and neck particle models 10l1, 10l2, 10l3, and 10l4 that display only the seven muscle types that have been dynamically optimized, and dynamic three-dimensional head and neck particle models 10m1, 10m2, 10m3, and 10m4 that hide the muscles and enlarge the area around the hyoid bone.
[0474] In Figure 41, the overall dynamic three-dimensional head and neck particle model 10k1, the partially displayed dynamic three-dimensional head and neck particle model 10l1, and the enlarged dynamic three-dimensional head and neck particle model 10m1 near the hyoid bone are shown in their states 0.133 S after the start of the swallowing simulation.
[0475] In addition, in Figure 41, the overall dynamic three-dimensional head and neck particle model 10k2, the partially displayed dynamic three-dimensional head and neck particle model 10l2, and the enlarged dynamic three-dimensional head and neck particle model 10m2 near the hyoid bone are shown in their states 1.133 S after the start of the swallowing simulation.
[0476] In Figure 42, the overall image of dynamic three-dimensional head and neck particle model 10k3, the partially displayed dynamic three-dimensional head and neck particle model 10l3, and the enlarged dynamic three-dimensional head and neck particle model 10m3 near the hyoid bone are shown 1,900 seconds after the start of the swallowing simulation.
[0477] In addition, in Figure 42, the overall dynamic three-dimensional head and neck particle model 10k4, the partially displayed dynamic three-dimensional head and neck particle model 10l4, and the enlarged dynamic three-dimensional head and neck particle model 10m4 near the hyoid bone are shown in their states 2,500 seconds after the start of the swallowing simulation.
[0478] From Figures 41 and 42, it was confirmed that even in the dynamic three-dimensional head and neck particle models 10k1 to 10k4 that faithfully reproduce the head and neck organs, the optimal muscle activity rate α can be set for each muscle type by manual setting processing, static optimization processing, and dynamic optimization processing.
[0479] It was confirmed that the dynamic three-dimensional head and neck particle models 10k1 to 10k4, which faithfully reproduce the head and neck organs, had less vibration in the muscle activity rate α than the simplified dynamic three-dimensional head and neck particle model 50a shown in Fig. 11. Therefore, it was found that the setting of the relaxation degree β was more effective in the simplified dynamic three-dimensional head and neck particle model 50a.
[0480] (9) <Action and Effects> In the above configuration, in the swallowing simulation device 1, a plurality of head and neck organs are modeled in a three-dimensional image using a plurality of particles, and a dynamic three-dimensional head and neck particle model created by a plurality of muscle particles in which the muscle fiber directions of the plurality of head and neck organs are specified within the three-dimensional image for each muscle type m is stored in the memory unit 83.
[0481] In the swallowing simulation device 1, a muscle activity rate α m The value of is changed by the muscle activity rate changing unit 80, and the movement of the head and neck organs during swallowing is analyzed in three-dimensional images based on the particle method using a dynamic three-dimensional head and neck particle model with a changed muscle activity rate.
[0482] In this embodiment, the muscle activity rate α set for each muscle type m of the contractile muscle of the head and neck organs is mThe motion of each muscle particle i of the dynamic three-dimensional head and neck particle model 10c, 50a during swallowing is analyzed in a three-dimensional image by reflecting the temporal change of the muscle activity rate α m k,Ph By setting the optimum value, it is possible to reproduce the contractile muscles that move actively over time during swallowing, and to reproduce the movement of the head and neck organs during swallowing more accurately than ever before.
[0483] In addition, in the swallowing simulation device 1, the motion analysis unit 50 calculates the muscle activity rate α m k,Ph The movement of the head and neck organs during swallowing was analyzed based on the particle method when the position of the muscle particles r i k,Ph Here, the muscle activity rate α m k,Ph The analysis of the movement of the head and neck organs during swallowing using a three-dimensional image based on the particle method using dynamic three-dimensional head and neck particle models 10c, 50a with the values of changed is one cycle of calculation processing SRa, SRb by the particle method as shown in Figures 9 and 10, and the position r of the muscle particle is calculated from the result of this calculation processing.
[0484] The swallowing simulation device 1 is a device for measuring the muscle activity rate α m k,Ph and the position r of the muscle particle obtained by the particle method analysis. i k,Ph Based on this, the evaluation function e k,Ph The swallowing simulation device 1 calculates the obtained evaluation function e k,Ph Based on this, the muscle activity rate during swallowing can be analyzed.
[0485] The swallowing simulation device 1 displays, as a moving image, the analysis results of the movement of the head and neck organs from the start of swallowing to the end of swallowing, which are analyzed using three-dimensional images by the movement analysis unit 50.
[0486] As a result, in the swallowing simulation device 1, the muscle activity rate α m k,Ph The evaluation function e calculated each time the value of k,Ph Based on this, the muscle activity rate α m k,Ph It is possible to determine the value of , and to reproduce the movement of the head and neck organs during swallowing more accurately than ever before.
[0487] In addition, the swallowing simulation device 1 performs a swallowing simulation to analyze the movement of the head and neck organs, and calculates the muscle activity rate α m k,Ph and the position r of the muscle particle obtained by the particle method analysis. i k,Ph And the velocity of muscle particles v i k,Ph Based on this, the evaluation function e k,Ph As a result, the swallowing simulation device 1 calculates the obtained evaluation function e k,Ph Based on this, the velocity of the muscle particle v i k,Ph Taking into account the muscle activity rate α during swallowing m k,Ph can be analyzed.
[0488] (10) <Other embodiments> In the above-described embodiment, in addition to forming the dynamic three-dimensional head and neck particle models 10c, 50a that reproduce the head and neck organs of a swallowing-disabled person or a healthy person using three-dimensional images, a dynamic three-dimensional head and neck particle model that reproduces the head and neck organs of, for example, an infant or an elderly person may be formed using three-dimensional images.
[0489] In the above-described embodiment, a particle method such as the SPH (Smoothed Particle Hydrodynamics) method may be applied in addition to the MPS method employed in the present embodiment.
[0490] In the above embodiment, the case where the formula (8) is applied has been described, but the present invention is not limited to this. For example, the correction coefficient f l may always be set to 1. In addition, for example, a correction coefficient fv according to the rate of change of muscle fiber length may be further added to the above formula (8) by multiplication.
[0491] In addition, in the dynamic three-dimensional head and neck particle model, muscle particles may be set for the muscle types of the head and neck organs, such as the levator pharyngeal muscles, tongue muscles, soft palate muscles, suprahyoid muscles, infrahyoid muscles, laryngeal muscles, etc., and the muscle particles may be moved based on the muscle activity rate specified in the three-dimensional image during swallowing. This allows the muscle activity rates of these muscle types to be optimized based on the evaluation function.
[0492] In addition, in the above-described embodiment, a first technique of performing manual setting processing and static optimization processing, and a second technique of performing manual setting processing, static optimization processing, and dynamic optimization processing have been described. However, the present invention is not limited to this. For example, dynamic optimization processing may be performed after manual setting processing, or static optimization processing and dynamic optimization processing may be performed without performing manual setting processing.
[0493] In the above-described embodiment, the passive stress S passive Using the method for calculating stress in a hyperelastic body of an isotropic Mooney-Rivlin material, the passive stress S is calculated based on the above equation (11). passive However, the present invention is not limited to this. passive As a method for calculating , for example, it is possible to apply the constitutive law of a hyperelastic body called an Ogden body instead of the Mooney-Rivlin body, or to apply the constitutive law of an anisotropic material based on muscle fibers in addition to the constitutive law of an isotropic material.
[0494] The computational processing (swallowing simulation) using the particle method described in the above "(7)<Muscle activity rate analysis processing>" to "(8)<Other verification results>" is described as being performed according to the above "(5-2)<Structural analysis method of dynamic three-dimensional head and neck particle model incorporating only internal forces>", but the present invention is not limited to this. In the above "(7)<Muscle activity rate analysis processing>", for example, as shown in Figures 2 and 3, it is also possible to perform computational processing according to the above "(5-1)<Structural analysis method of dynamic three-dimensional head and neck particle model incorporating internal and external forces>" using a dynamic three-dimensional head and neck particle model 10c that faithfully reproduces the head and neck organs and a pseudo-oral ingestible product modeled by particles.
[0495] When performing calculation processing according to the above-mentioned "(5-1) <Structural analysis method for dynamic three-dimensional head and neck particle model incorporating internal and external forces>", the "one cycle of calculation processing SRa using particle method" from steps S1 to S9 shown in Figure 9 can be applied as the "one cycle of calculation processing using particle method" in subroutine SR5 of Figure 15, subroutines SR6 and SR7 of Figure 17A, subroutine SR8 of Figure 17B, subroutines SR11 and SR12 of Figure 22A, subroutine SR13 of Figure 22B, and subroutines SR16, SR17, and SR18 of Figure 27.
[0496] (11) Embodiments using other evaluation functions (11-1) Overview In the above-described embodiment, the evaluation function e used in the dynamic optimization process k,Ph Although the above formula (35) is used as the general formula for , the present invention is not limited to this, and the following formula (37) may also be applied.
number
[0497] In the above equation (37), e k,Phindicates an evaluation function. The superscript k indicates the number of optimization calculations of the evaluation function calculated by muscle activity rate analysis unit 91, and the superscript Ph indicates an identifier for identifying the change state of muscle activity rate when the evaluation function is calculated while changing the value of muscle activity rate for each muscle type. Os represents the assembly of multiple head and neck organs in a dynamic 3D head and neck particle model. Oj denotes a head and neck organ for which an evaluation function is calculated, among a plurality of head and neck organs. w V,Oj indicates a constant for weighting the head and neck organs (Oj) for which the evaluation function is calculated. v i k,Ph is the analysis result by the particle method in the motion analysis unit 50 obtained when calculating the evaluation function for the kth optimization calculation, and indicates the velocity of muscle particle i in the three-dimensional image at a given time during swallowing. u i * is the target velocity of muscle particle i, and the subscript i is an identifier that identifies muscle particle i. i * As shown in the following equation (38), the reference speed u - i and the target speed u i It is determined by the weighted average of w α indicates a constant for weighting the muscle activity rate. α m k,Ph indicates the muscle activity rate of muscle particle i in muscle type m used when calculating the evaluation function for the kth optimization calculation. The subscript m is an identifier defined for each muscle type of head and neck organs.
[0498] In the above embodiment, the target velocity u of the muscle particle i is i * The case where the target velocity u of muscle particle i is calculated based on the following equation (38) will be described, but the present invention is not limited to this. i * may be calculated from various calculation expressions.
number
[0499] u i denotes the target velocity of muscle particle i in the 3D image at a given time during swallowing. i is the velocity of muscle particle i when the muscle particle i is moved so as to match a 4DCT image (analysis image used to analyze the swallowing of the subject) obtained when the subject swallows, for example. u - i is the reference velocity of muscle particle i, and is calculated using the following equation (39). w u- is the reference velocity of muscle particle i, u - i and u, the target velocity of muscle particle i in the 3D image. i and w is a weighting constant for u- is 0 <w u- <1, but from the perspective of reproducing a more natural swallowing movement in the dynamic three-dimensional head and neck particle model, 0.05≦w u- Analysis using swallowing simulation has confirmed that values in the range of ≦0.3 are particularly desirable.
number
[0500] s i indicates the target position of muscle particle i in the 3D image at a given time during swallowing. i is, for example, the position (coordinate) of muscle particle i when muscle particle i is moved so as to match a 4DCT image obtained when a subject swallows. △t opt indicates the time intervals from the start of swallowing to the end of swallowing, and the time interval from the start of swallowing to the end of swallowing is time △t opt This is a periodic division. opt is also called the control period. r i tsThe dynamic three-dimensional particle model of the head and neck is used to analyze the movement of the head and neck organs using three-dimensional images based on the particle method. The muscle activity rate is determined using an evaluation function from the time from the start of swallowing to the end of swallowing, and the optimization target time ts~(ts+△t opt ) is set, the optimization target time ts~(ts+△t opt ) and shows the position of muscle particle i in the three-dimensional image at the optimization start time ts. v i ts is a dynamic three-dimensional particle model of the head and neck that analyzes the movement of the head and neck organs using three-dimensional images based on the particle method. The muscle activity rate is determined using an evaluation function from the time from the start of swallowing to the end of swallowing. The optimization target time ts~(ts+△t opt ) is set, the optimization target time ts~(ts+△t opt ) and shows the velocity of muscle particle i in the three-dimensional image at the optimization start time ts.
[0501] The above equation (39) is derived from the following equation (40).
number
[0502] The above equation (40) is expressed as the coordinate r at the optimization start time ts. i ts The muscle particle i located at time △t opt Then, coordinate s i The speed of the object is then approximated by u - i and v i ts The average of these is used.
[0503] The first term on the right-hand side of the above equation (37) indicates the error in the position and velocity of muscle particle i that occurred in the particle method analysis of the 4DCT image obtained when the subject swallowed. The second term on the right-hand side of the above equation (37) is a term provided to suppress the activity of muscles unrelated to the movement of muscle particle i and the activity of antagonistic muscles. The ratio of these terms is expressed by the weighting constant w V,Oj and w α is determined, but for example, w V,Oj =1 or w α =10 -6 etc., and make appropriate adjustments.
[0504] Here, we present a dynamic 3D head and neck particle model that uses a 3D particle-based method to reproduce a more natural swallowing motion. V,Oj The value of 10 -2 ~10 2 Set the range to w α The value of is 0~w V,Oj It is desirable to set it in the range of
[0505] In addition, w in the above equation (38) u- defines the ratio of the coordinate error (the first term) to the velocity error (the second term) in equation (38), and w u- The larger the value of , the greater the feedback given to the coordinate error.
[0506] In this case, the swallowing simulation device 1 calculates the muscle activity rate α m t1 Only the muscle activity rate α at time t2, t3, ... is calculated by static optimization processing. m t2 ,α m t3 , ... should be calculated according to FIG. 25. However, in this embodiment, Δt opt is set to, for example, 0.1S or more, making the relaxation process unnecessary. It is preferable to perform the analysis process shown in FIGS. 22A and 22B as the analysis process of the subroutine SR15 in FIG. 25.
[0507] More specifically, in the above-mentioned "(7-3) <Dynamic optimization process of another embodiment in which relaxation process is performed>", Δt opt For example, the extremely small time is set to 0.00001 to 0.000500 S. Here, △t opt It is desirable to set the time between 0.01S and 0.2S. opt If it is set to less than 0.1 S, it will be impossible to take into account the activity of muscles away from the head and neck organs for which the evaluation function is calculated, whereas if it is set to more than 0.2 S, the control period will become too coarse and it will not be possible to reproduce the swallowing movement, so it is desirable to set it to between 0.01 S and 0.2 S.
[0508] In this way, the time Δt opt As a result, as shown in Figure 23, the optimization target time ts~(ts+△t opt ) and muscle activity rate α at the first calculation time (t+△t) k,p(t+△t) is identified and set as the muscle activity rate used in one cycle of calculation processing using the particle method. After that, the calculation time (t+△t) is the optimization end time (ts+△t opt 22B, the processes of step S92, subroutine SR13, and steps S93 to S95 shown in FIG. 22B are repeated until the process reaches the state where the number of the first subroutines is 100.
[0509] In this embodiment, the subroutine SR15 in the flowchart of FIG. 25 is replaced with the analysis process shown in FIG. 22A and FIG. 22B, and the relaxation process is not required. Therefore, the muscle activity rate α kmax,pend(β) is muscle activity rate α kmax,pend In addition, in this case, γ·α is set as the initial solution in step S111 of FIG. ts It is desirable to do so.
[0510] It is preferable that γ is 0<γ<1, and more preferably 0.5≦γ≦0.75. If γ is set to a value close to 0, there is a possibility that an appropriate solution will not be reached by the maximum number kmax of optimization calculations k, so it is preferable to set γ<0. On the other hand, if γ is set to a value close to 1, a problem may occur in which the activity rate does not decrease even after swallowing in muscles that should decrease in activity rate after swallowing during swallowing simulation using a dynamic three-dimensional head and neck particle model, so it is preferable to set γ<1.
[0511] Note that detailed explanations of the dynamic optimization processing procedure shown in FIG. 25 and the analysis processing used therein in FIGS. 22A and 22B will be omitted here to avoid duplication, but the outline of the processing is as follows.
[0512] In this case, the initial solution is γ α ts After setting, the gradient ▽e k Then, the search direction ya k Here, we also perform particle method analysis from the start time to the end time of swallowing for the number of muscle types m, and calculate (∂e / ∂α m ) k Then, the above-mentioned line search is performed to update the solution. p Gradually increase and perform particle method analysis from the start time to the end time of swallowing each time, and calculate the evaluation function e from the above equation (37). k,Ph Then, this evaluation function e k,Ph The step size μ at which p Calculate the gradient ▽e k The calculation of and line search processing are performed using the evaluation function e k,Ph When it stops decreasing or when the process is repeated a pre-specified number of times, a local optimum solution for the muscle activity rate is obtained.
[0513] In the above configuration, the muscle activity rate analysis unit 91 according to this embodiment calculates an evaluation function e from the above equation (37) based on the position and velocity of the muscle particle i in the three-dimensional image at a predetermined time during swallowing, which are the analysis results by the particle method in the motion analysis unit 50, the target position and target velocity of the muscle particle i in the three-dimensional image at a predetermined time during swallowing, and the muscle activity rate α used when the motion of the muscle particle i was analyzed by the motion analysis unit 50 using the particle method. k,Ph Then, the muscle activity rate analysis unit 91 calculates the obtained evaluation function e k,Ph Based on the calculation result, muscle activity rate α m k,Ph We will identify the following.
[0514] As a result, the swallowing simulation device 1 according to the other embodiment also m k,Ph can be set to an optimal value. Therefore, the swallowing simulation device 1 can reproduce the contractile muscles that move actively with the passage of time during swallowing, and can reproduce the movement of the head and neck organs during swallowing more accurately than ever before.
[0515] In this other embodiment, the dynamic optimization process of FIG. 25 is used, and the analysis process shown in FIG. 22A and FIG. 22B is performed as the analysis process of subroutine SR15 of FIG. 25. The above-mentioned equation (37) is used as the evaluation function used in this process. However, the present invention is not limited to this, and the above-mentioned equation (35) may be used instead of equation (37).
[0516] (11-2) Verification test Here, the weighting constant w u-A verification test was conducted to examine the trajectory of the center of gravity of the hyoid bone 52 (FIG. 11) during a swallowing simulation by changing the numerical value of . As a result, the results shown in FIGS. 43A to 43C were obtained. Here, the dynamic optimization process shown in FIG. 25 was performed using the simplified dynamic three-dimensional head and neck particle model 50a shown in FIG. 11. However, here, as described above, the analysis process shown in FIGS. 22A and 22B was performed without performing the relaxation process as subroutine SR15. The time step Δt in the particle method analysis was set to Δt=5×10 -5 [s], and the control period of the muscle activity rate for the optimization calculation is △t opt △t opt =0.1[s].
[0517] Weighting constant w u- The value of was set to 0.0, 0.1, and 0.5, and the behavior of the center of gravity of the hyoid bone 52 (FIG. 11) during each swallowing simulation was confirmed, and the results shown in FIGS. 43A to 43C were obtained. In FIGS. 43A to 43C, the horizontal axis indicates the position on the y-axis, which is the front-back direction of the body parallel to the median plane of the dynamic three-dimensional head and neck particle model 50a. The vertical axis indicates the position on the z-axis, which is the up-down direction of the body of the dynamic three-dimensional head and neck particle model 50a. Note that "4DCT" in FIGS. 43A to 43C indicates the target value, and is obtained by tracing the center of gravity of the subject's hyoid bone based on the 4DCT image when the subject swallows.
[0518] w u- When w = 0.0, the trajectory of the hyoid bone 52 was roughly similar to the target value, but since the error was not taken into account in the evaluation function of the optimization analysis, it accumulated and the coordinates remained shifted downward after time 1.5 [s]. u- When =0.5, the feedback to the error in the coordinates of the center of gravity of the hyoid bone 52 was too strong, causing an overshoot, and as a result, the hyoid bone 52 exhibited a vibrating behavior as shown in FIG. 43C.
[0519] w u- When = 0.1, the average error (deviation from the target value) during the swallowing simulation was 1.3 mm, and it was confirmed that the movement of the hyoid bone 52 could be reproduced well. <wu- In the range <1, especially 0.0 <w u- <0.5 range, and even 0.05≦w u- It was confirmed that a range of ≦0.3 is desirable.
[0520] Next, the value of γ set as the initial value was set to 1.0, 0.75, and 0.5, and the behavior of the center of gravity of the hyoid bone 52 (FIG. 11) during each swallowing simulation was confirmed, and the results shown in FIG 44 were obtained. As shown in FIG 44, in the simplified dynamic three-dimensional head and neck particle model 50a, no significant difference in the behavior of the hyoid bone 52 was observed even when the value of γ was changed.
[0521] Next, the initial value of γ was set to 1.0, 0.75, and 0.5, and the muscle activity rates of the stylohyoid muscles 56a, 56b, geniohyoid muscle 57, and sternohyoid muscle 58 during each swallowing simulation were calculated, resulting in the results shown in Figures 45A to 45C.
[0522] In Figures 45A to 45C, the change over time in the muscle activity rate of the stylohyoid muscles 56a and 56b is indicated by StyH, the change over time in the muscle activity rate of the geniohyoid muscle 57 is indicated by GH, and the change over time in the muscle activity rate of the sternohyoid muscle 58 is indicated by SteH.
[0523] When the initial value was not decreased, γ=1.0, the antagonistic muscles continued to exert muscle force after swallowing (Time=2.5[s] or later) in the simplified dynamic three-dimensional head and neck particle model 50a, resulting in an unnatural swallowing motion. On the other hand, when the initial value was decreased, γ=0.75 or 0.5, the muscle force decreased after swallowing (Time=2.5[s] or later) in the simplified dynamic three-dimensional head and neck particle model 50a, resulting in a good swallowing motion. From the above, it was confirmed that γ is preferably 0<γ<1, and more preferably 0.5≦γ≦0.75. [Explanation of symbols]
[0524] 1. Swallowing simulation device 4 Display 10c,50a Dynamic 3D head and neck particle model 10 Head and Neck Modeling Section 30 Organ movement setting section 31 Forced movement setting section 32 Muscle contraction motion setting section 50 Motion Analysis Department 80 Muscle activity rate change unit 91 Muscle activity rate analysis section
Claims
1. A storage unit that stores a dynamic three-dimensional head and neck particle model in which a plurality of head and neck organs are modeled by a plurality of muscle particles in which muscle fiber directions are specified in a three-dimensional image for each muscle type, and a contraction stress based on the muscle fiber directions during swallowing is applied to the muscle particles according to a muscle activity rate; A muscle activity rate changing unit that changes the value of the muscle activity rate set for each muscle type; a motion analysis unit that uses the dynamic three-dimensional head and neck particle model in which the value of the muscle activity rate has been changed to analyze the motion of the head and neck organs during swallowing using the three-dimensional image based on a particle method; a muscle activity rate analysis unit that calculates an evaluation function for evaluating the muscle activity rate based on an analysis result of the movement of the head and neck organs during swallowing, which is analyzed by the movement analysis unit using a particle method, and analyzes the muscle activity rate based on the evaluation function; A swallowing simulation device comprising:
2. The muscle activity rate analysis unit is the position of the muscle particle in the three-dimensional image at a predetermined time during swallowing, which is an analysis result by the particle method in the motion analysis unit; and a target position of the muscle particle in the three-dimensional image at the predetermined time during swallowing; The muscle activity rate used when the motion of the muscle particles is analyzed by the motion analysis unit using a particle method; and The swallowing simulation device according to claim 1 , wherein the evaluation function is calculated based on the following:
3. The muscle activity rate analysis unit calculates the evaluation function based on the following formula (1): The swallowing simulation device according to claim 1 . [0010] …(1) e k,Ph The superscript k indicates the number of optimization calculations of the evaluation function calculated by the muscle activity rate analysis unit, and the superscript Ph indicates an identifier for identifying a change state of the muscle activity rate when the evaluation function is calculated while changing the value of the muscle activity rate for each muscle type. Os represents a collection of the plurality of head and neck organs in the dynamic 3D head and neck particle model. Oj indicates the head and neck organ for which the evaluation function is calculated, among the plurality of head and neck organs. W P,Oj indicates a constant for weighting the head and neck organs (Oj) for which the evaluation function is calculated. r i k,Ph is the analysis result by the particle method in the motion analysis unit obtained when calculating the evaluation function for the kth optimization calculation, and indicates the position of muscle particle i in the three-dimensional image at a predetermined time during swallowing. The subscript i is an identifier that identifies the muscle particle i. s i indicates a target position of the muscle particle i in the three-dimensional image, which is a target at the predetermined time during swallowing. W α indicates a constant for weighting the muscle activity rate. α m k,Ph indicates the muscle activity rate of the muscle particle i in the muscle type m used when calculating the evaluation function for the kth optimization calculation. The subscript m is an identifier defined for each muscle type of the head and neck organs.
4. The muscle activity rate analysis unit is The position and velocity of the muscle particles in the three-dimensional image at a predetermined time during swallowing, which are analysis results using a particle method in the motion analysis unit; a target position and a target velocity of the muscle particle in the three-dimensional image at the predetermined time during swallowing; The muscle activity rate used when the motion of the muscle particles is analyzed by the motion analysis unit using a particle method; and The swallowing simulation device according to claim 1 , wherein the evaluation function is calculated based on the following:
5. The muscle activity rate analysis unit calculates the evaluation function based on the following formula (2): The swallowing simulation device according to claim 1 . [0025] …(2) e k,Ph The superscript k indicates the number of optimization calculations of the evaluation function calculated by the muscle activity rate analysis unit, and the superscript Ph indicates an identifier for identifying a change state of the muscle activity rate when the evaluation function is calculated while changing the value of the muscle activity rate for each muscle type. Os represents a collection of the plurality of head and neck organs in the dynamic 3D head and neck particle model. Oj indicates the head and neck organ for which the evaluation function is calculated, among the plurality of head and neck organs. W P,Oj and w V,Oj indicates a constant for weighting the head and neck organs (Oj) for which the evaluation function is calculated. r i k,Ph is the analysis result by the particle method in the motion analysis unit obtained when calculating the evaluation function for the kth optimization calculation, and indicates the position of muscle particle i in the three-dimensional image at a predetermined time during swallowing. The subscript i is an identifier that identifies the muscle particle i. s i indicates a target position of the muscle particle i in the three-dimensional image, which is a target at the predetermined time during swallowing. v i k,Ph is the analysis result using the particle method in the motion analysis unit obtained when calculating the evaluation function for the kth optimization calculation, and indicates the velocity of the muscle particle i in the three-dimensional image at the specified time during swallowing. u i indicates a target velocity of the muscle particle i in the three-dimensional image at the predetermined time during swallowing. W α indicates a constant for weighting the muscle activity rate. α m k,Ph indicates the muscle activity rate of the muscle particle i in the muscle type m used when calculating the evaluation function for the kth optimization calculation. The subscript m is an identifier defined for each muscle type of the head and neck organs.
6. The muscle activity rate analysis unit is The dynamic three-dimensional head and neck particle model analyzes the movement of the head and neck organs using the three-dimensional image based on the particle method. The optimization target time ts to (ts + Δt opt ) and The optimization target time ts to (ts+Δt opt Optimization end time (ts+Δt opt ) Muscle activation rate α kmax,pend is determined based on the evaluation function, The muscle activity rate α at the optimization start time ts from the optimization target time ts to (ts+Δt opt ) ts and the optimization end time (ts+Δt opt ) the muscle activity rate α kmax,pend Using the relaxation degree β, the muscle activity rate α after relaxation is calculated by the following formula (3). kmax,pend(β) The swallowing simulation device according to claim 1 , wherein the swallowing simulation device calculates: [0030] …(3) α ts indicates the value of the muscle activity rate at the optimization start time ts. α kmax,pend indicates the value of the muscle activity rate finally determined based on the evaluation function before the relaxation process is performed. β is the degree of relaxation and indicates a predetermined constant. α kmax,pend(β) is the muscle activity rate for which the relaxation process has been performed, and the superscript (β) is an identifier indicating that the relaxation process has been performed.
7. The muscle activity rate analysis unit calculates the evaluation function based on the following formula (4): The swallowing simulation device according to claim 1 . [0045] …(4) e k,Ph The superscript k indicates the number of optimization calculations of the evaluation function calculated by the muscle activity rate analysis unit, and the superscript Ph indicates an identifier for identifying a change state of the muscle activity rate when the evaluation function is calculated while changing the value of the muscle activity rate for each muscle type. Os represents a collection of the plurality of head and neck organs in the dynamic 3D head and neck particle model. Oj indicates the head and neck organ for which the evaluation function is calculated, among the plurality of head and neck organs. W V,Oj indicates a constant for weighting the head and neck organs (Oj) for which the evaluation function is calculated. v i k,Ph is the analysis result using the particle method in the motion analysis unit obtained when calculating the evaluation function for the kth optimization calculation, and indicates the velocity of the muscle particle i in the three-dimensional image at a specified time during swallowing. u i * is the target velocity of the muscle particle i, and the subscript i is an identifier that identifies the muscle particle i. W α indicates a constant for weighting the muscle activity rate. α m k,Ph indicates the muscle activity rate of the muscle particle i in the muscle type m used when calculating the evaluation function for the kth optimization calculation. The subscript m is an identifier defined for each muscle type of the head and neck organs.
8. Said u i * is calculated from the following formula (5): The swallowing simulation device according to claim 7. [0050] …(5) u i indicates a target velocity of the muscle particle i in the three-dimensional image at the predetermined time during swallowing. u - i is the reference velocity of the muscle particle i, and is calculated using the following equation (6). W u- is the reference velocity of the muscle particle i, u - i and u, the target velocity of the muscle particle i in the three-dimensional image. i and a weighting constant, 0 < w u- It is a number determined in the range of <1. [006] …(6) s i indicates a target position of the muscle particle i in the three-dimensional image, which is a target at the predetermined time during swallowing. △t opt indicates the time that divides the period from the start of swallowing to the end of swallowing into predetermined intervals. r i ts is a time period from the start of swallowing to the end of swallowing in which the movement of the head and neck organs is analyzed in the three-dimensional image based on the particle method using the dynamic three-dimensional head and neck particle model, and the muscle activity rate is specified using the evaluation function from among the times from the start of swallowing to the end of swallowing. opt ) is set, the optimization target time ts to (ts + Δt opt ) is an analysis result by the particle method in the motion analysis unit, and shows the position of muscle particle i in the three-dimensional image at the optimization start time ts. v i ts The dynamic three-dimensional head and neck particle model analyzes the movement of the head and neck organs using the three-dimensional image based on the particle method. The optimization target time ts to (ts + Δt opt ) is set, the optimization target time ts to (ts + Δt opt ) is an analysis result using the particle method in the motion analysis unit, and shows the velocity of the muscle particle i in the three-dimensional image at the optimization start time ts.
9. The swallowing simulation device according to any one of claims 1 to 8, wherein each time the muscle activity rate changing unit changes a value of the muscle activity rate for each muscle type, the motion analysis unit analyzes the motion of the muscle particles using a particle method, the muscle activity rate analysis unit calculates the evaluation function for each analysis result by the particle method in the motion analysis unit, and the muscle activity rate is determined based on a plurality of evaluation functions.
10. The muscle activity rate analysis unit is After calculating the evaluation function, the step size μ p While changing the value of m k,p The muscle activity rate α m k,p The swallowing simulation device according to any one of claims 1 to 9, wherein the motion of the muscle particles is analyzed by a particle method using a particle method, and the evaluation function is calculated for each analysis result by the particle method in the motion analysis unit. [0070] …(7) α m k,p indicates the muscle activity rate of the muscle particle i in muscle type m used when calculating the evaluation function for the kth optimization calculation. The subscript m is an identifier defined for each muscle type of the head and neck organ. p indicates an identifier for identifying that the muscle activity rate is in a predetermined changed state. α m k,0 indicates an initial muscle activity rate of the muscle particle i in the muscle type m of the head and neck organ used when calculating the evaluation function for the kth optimization calculation. μ p indicates a changeable value. ya m k indicates the search direction determined based on the evaluation function.
11. The motion analysis unit is Analyzing at least the elastic force applied to the muscle particle, the artificial potential force applied to the muscle particle, and the viscous force applied to the muscle particle by the particle method; When analyzing the elastic force, the second Piola-Kirchhoff stress tensor expressed by the following formula (8) is included in the analysis as the contractile stress applied to the muscle particle. The swallowing simulation device according to any one of claims 1 to 10. [0080] …(8) S passive represents the passive stress, which is the passive stress experienced due to the movement of other particles. S active represents the active contraction stress, which is the active contraction stress based on the muscle fiber direction, and is calculated by an arithmetic formula including the muscle activity rate indicating the temporal change in the contraction stress of muscle particle i in muscle type m during swallowing in the dynamic three-dimensional head and neck particle model, and the muscle fiber direction at the initial time 0 set for muscle particle i of muscle type m.
12. a storing step of storing a dynamic three-dimensional head and neck particle model in which a plurality of head and neck organs are modeled by a plurality of muscle particles in which muscle fiber directions are specified in a three-dimensional image for each muscle type, and a contraction stress based on the muscle fiber directions during swallowing is applied to the muscle particles by a muscle activity rate; A muscle activity rate changing step of changing the value of the muscle activity rate set for each muscle type; a motion analysis step of analyzing the motion of the head and neck organs during swallowing based on a particle method using the dynamic three-dimensional head and neck particle model in which the value of the muscle activity rate has been changed, using the three-dimensional image; a muscle activity rate analysis step of calculating an evaluation function for evaluating the muscle activity rate based on an analysis result of the movement of the head and neck organs during swallowing, which is analyzed by a particle method in the movement analysis step, and analyzing the muscle activity rate based on the evaluation function; A swallowing simulation method comprising:
Citation Information
Patent Citations
Food product development assistance device, food product development method, food product production method, dietary education assistance device, and dietary education method
CN103597510A
Magnetic recording and reproducing system
JP1985022789A
Deglutition simulation apparatus and method
JP2013202119A
Swallowing Simulation Apparatus and Method
US20140120509A1
